Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
2016 Health Home Evaluation Report
June 14, 2017
2016 Health Home Evaluation Report
Table of Contents
Executive Summary ..........................................................................................................................i
Introduction ......................................................................................................................................1
Section 1. The Health Home Model ..................................................................................................1
Background .................................................................................................................................1
Health Home Programs Nationwide ..........................................................................................2
Section 2. The Maryland Health Home Program .............................................................................4
Health Home Data .......................................................................................................................6
Participant Characteristics..........................................................................................................6
Health Home Services .................................................................................................................9
Section 3. Health Home Participants Health Care Utilization Patterns .........................................15
Composition of Total Medicaid Services ...................................................................................15
Emergency Department Utilization Patterns ............................................................................16
Health Care Utilization by Length of Enrollment ......................................................................17
Section 4. Health Home Participant Outcomes ..............................................................................19
Evaluation Cohort Description ...................................................................................................19
Demographic Characteristics .....................................................................................................20
Health Care Utilization Outcomes .............................................................................................22
Inpatient Hospital Admissions .............................................................................................22
Lengths of Inpatient Hospital Stay ......................................................................................23
Emergency Department Utilization .....................................................................................25
Ambulatory Care Utilization .................................................................................................25
Nursing Home Admissions ...................................................................................................26
Health Care Quality Outcomes ...................................................................................................27
30-Day All-Cause Hospital Readmissions .............................................................................27
Appropriateness of ED Care .................................................................................................28
Health Care Cost Outcomes .......................................................................................................30
Section 5. Health Home Regression Analysis Results .....................................................................36
Emergency Department Utilization .....................................................................................36
Inpatient Hospital Admissions .............................................................................................38
Non-Emergent Emergency Department Visits ....................................................................39
Avoidable Hospital Admissions ............................................................................................41
30-Day Hospital Readmissions .............................................................................................42
Cost Outcomes ......................................................................................................................44
Limitations .........................................................................................................................................45
Conclusion .........................................................................................................................................47
Appendix A: Evaluation Cohort Selection .......................................................................................36
Appendix B: Regression Methods ...................................................................................................49
Model Selection ..........................................................................................................................49
Interpretation of Tables .............................................................................................................50
References ........................................................................................................................................52
List of Tables and Figures Tables
Table 1. Demographic and Clinical Characteristics of Health Home Participants ..........................9
Table 2. Number and Percentage of Health Home Participants with More than One Body Mass Index (BMI) Recorded, According to HEDIS BMI Thresholds ...............................................14
Table 3. Number and Percentage of Health Home Participants with More than One Blood Pressure (BP) Recorded, According to HEDIS BP Thresholds ........................................................14
Table 4. Percentage of Health Home Participants Completing ED Visits, by Service Type, CY 2013 – CY 2015 ....................................................................................................................................16
Table 5. ED Utilization Rates per Participant, by Length of Enrollment ........................................18
Table 6. Inpatient Utilization Rates per Participant, by Length of Enrollment .............................18
Table 7. Demographic and Clinical Characteristics of Study Group and Comparison Group .......21
Table 8. Program Types of the Study Group and Comparison Group ...........................................22
Table 9. Percentage of Participants with at Least One Inpatient Hospital Admission, by Treatment Group, CY 2013 – CY 2015 ...............................................................................................23
Table 10. Average Length of Stay for Inpatient Hospital Admission, by Treatment Group, CY 2013 – CY 2015 ....................................................................................................................................24
Table 11. Percentage of Participants with at Least One Emergency Department Visit, by Treatment Group, CY 2013 – CY 2015 ...............................................................................................25
Table 12. Percentage of Participants with at Least One Ambulatory Care Visit, by Treatment Group, CY 2013 – CY 2015 ..................................................................................................................26
Table 13. Percentage of Participants with at Least One Nursing Home Admission, by Treatment Group, CY2013 - CY2015 ..................................................................................................27
Table 14. Percentage of Participant with at Least One 30-Day All-Cause-Hospital Readmission, by Treatment Group, CY2013 - CY2015 .............................................................................................28
Table 15. Percentage of Participants with at Least One Non-Emergent ED Visit, by Treatment Group, CY2013 - CY2015 .....................................................................................................................29
Table 16. Average Hospital Inpatient Health Care Costs by Treatment Group and Provider Type, CY 2013‐ CY 2015 ......................................................................................................................31
Table 17. Average Outpatient Health Care Costs by Treatment Group and Provider Type, CY 2013‐ CY 2015 ......................................................................................................................................32
Table 18. Average Professional Services Health Care Costs by Treatment Group and Provider Type, CY 2013‐ CY 2015 ......................................................................................................................33
Table 19. Average Other Services Health Care Costs by Treatment Group and Provider Type, CY 2013‐ CY 2015 ................................................................................................................................34
Table 20. Average Total Health Care Costs per Person by Treatment Group and Provider Type, CY 2013‐ CY 2015 ......................................................................................................................35
Table 21. Results of Regression on Counts of ED Visits, CY 2013 and CY 2015 ...............................37
Table 22. Results of Regression on Counts of Inpatient Hospital Admissions, CY 2013 and CY 2015 ....................................................................................................................................................39
Table 23. Results of Regression on Counts of Non-Emergent ED Visits, CY 2013 and CY 2015 .....40
Table 24. Results of Regression on the Likelihood of an Avoidable Hospital Admission, CY 2013 and CY 2015 ...............................................................................................................................42
Table 25. Results of Regression on the Likelihood of a 30-day Hospital Readmission, CY 2013 and CY 2015 ........................................................................................................................................43
Table 26. Results of Regression on Total Annual Health Care Costs, CY 2013 and CY 2015 ..........45
Table 27. Number of Health Home and Comparison Group Participants ......................................48
Figures
Figure 1a. Number of Participating Health Home Providers, by Month ........................................5
Figure 1b. Number of Participating Health Home Provider Sites, by Month .................................5
Figure 2. Number of Health Home Participants, by Enrollee Type and Quarter ...........................7
Figure 3. Number of Health Home Participants, by Program Type and Quarter ..........................8
Figure 4. Percentage of Health Home Participants Receiving 0, 1, or 2 or More Services, by Month ................................................................................................................................................11
Figure 5. Average Number of Services Received by Health Home Participants, by Quarter .......12
Figure 6. Percentage of Health Home Participants by Types of Health Home Services Received, by Quarter ........................................................................................................................13
Figure 7. Composition of Types of Services Received by Health Home Participants, by Calendar Year ....................................................................................................................................15
Figure 8. Number and Composition of ED Visits Received by Health Home Participants, CY 2013 – CY 2015 ....................................................................................................................................17
Figure 9. Average Number of ED Visits, by Treatment Group, CY 2013 and CY 2015 ....................37
Figure 10. Average Number of Inpatient Hospital Admissions, by Treatment Group, CY 2013 and CY 2015 ........................................................................................................................................38
Figure 11. Average Number of Non-Emergent ED Visits, by Treatment Group, CY 2013 and CY 2015 ....................................................................................................................................................40
Figure 12. Percentage of Participants with at Least One Avoidable Hospital Admission, by Treatment Group, CY 2013 and CY 2015 ...........................................................................................41
Figure 13. Percentage of Participants with at Least One 30-Day Hospital Readmission, by Treatment Group, CY 2013 and CY 2015 ...........................................................................................43
Figure 14. Average Total Annual Health Care Costs, by Treatment Group, CY 2013 and CY 2015 .44
i
2016 Health Home Evaluation Report
Executive Summary
The Affordable Care Act (ACA) of 20101 presented an opportunity for states to improve care
coordination for Medicaid participants with chronic conditions by providing care through the
Health Home model. Under this legislation, each state can develop a program that offers a
person-centered approach to providing enhanced care management and care coordination. The
Maryland Department of Health and Mental Hygiene (DHMH) responded to this initiative and
submitted a Medicaid State Plan Amendment (SPA) that was approved by the Centers for
Medicare & Medicaid Services (CMS) in October 2013.
This report is an update of the 2015 Joint Chairmen’s Report on Patient Outcomes for
Participants in Health Homes. Its purpose is to describe the outcomes of participants in the
Maryland Health Home program. Maryland’s Health Home program targets Medicaid
participants with a serious and persistent mental illness (SPMI) and/or an opioid substance use
disorder (SUD) and risk of additional chronic conditions due to tobacco, alcohol, or other non-
opioid substance use and children with serious emotional disturbances (SED). Individuals can
participate in an Health Home if they are eligible for and engaged with a psychiatric
rehabilitation program (PRP), mobile treatment service (MTS), or an opioid treatment program
(OTP) that has been approved by DHMH to function as a Health Home provider.
Participating Health Homes receive an initial intake and assessment fee of $100.85 when they
enroll a new individual into the program. Health Home providers are also eligible for a $100.85
monthly rate per participant for each month in which an enrollee receives at least two qualified
Health Home services.2 If an enrollee receives fewer than two services, the Health Home is not
eligible to receive a payment for that individual for that month. Health Home services include
care coordination, care management, health promotion, and referrals to community and social
support services. The State received a 90% enhanced Federal Medical Assistance Percentage
(FMAP) for the provision of Health Home services during the first eight quarters of the program.
As of December 2016, payments to Health Home providers total approximately $10,187,000.
Since the inception of the program, over 7,900 participants have received services from nearly 40
Health Home providers in 67 individual sites across the state of Maryland. The majority of the
1 Pub. L. 111-148 (Mar. 23, 2010), as amended by the Health Care and Education Reconciliation Act of 2010, Pub.
L. 111-152 (Mar. 30, 2010). 2 Previous reports and presentations by the Department have referred to this payment as a ―per member per month
(PMPM)‖ payment. Since receipt of the monthly payment not guaranteed and is contingent on the provision of at
least two health home services by the enrollee, the characterization of the payment as a PMPM is not strictly
accurate. Program staff is in the process of updating the State’s SPA, regulations and related documents to reflect
this nuance.
ii
participants were between the ages of 18 and 64 years, resided in the Baltimore metropolitan
area, and were categorized as having moderate to very high co-morbidity levels. The most
frequently accessed provider type were PRP providers, which enrolled nearly 75 percent of all
Health Home participants.
The goal of the Health Home program is to improve health outcomes for individuals with
chronic conditions by providing patients with an enhanced level of care management and care
coordination while reducing costs. This evaluation is structured to provide a summary of health
care utilization, quality, and costs for calendar years (CYs) 2013 through 2015. The outcomes of
Health Home participants are compared with a comparison group of other Medicaid participants
with similar characteristics.
This analysis suggests that incremental progress towards achieving these goals may be
underway. However, preliminary results should be interpreted with caution, as sufficient time
has not passed since the implementation of the program to detect meaningful and sustained
differences in long-term health outcomes, as well as other factors such as small sample sizes and
limited data availability. Given these considerations, the results of these analyses suggest the
following:
Participation in a Health Home may be associated with an increase in the use of
ambulatory care services. Health Home study group participants with an ambulatory
care visit increased by 1.9 percentage points from 84.1 percent to 86.0 percent. During
the same time period, participants in the comparison group with at least one
ambulatory care visit decreased by 0.2 percentage points from 84.5 percent to 84.3
percent.
The descriptive analysis shows that the percentages of participants that had at least one
ED and/or one inpatient visit both decreased the longer those participants stayed in the
Health Home program. ED utilization rates decreased from 37.6 percent of participants
having at least one ED visit during the first six months of program participation to
27.5 percent with at least one ED visit during the 19 to 24 month enrollment span.
Inpatient utilization rates went from 10.0 percent of participants with at least one
inpatient visit during first six months of enrollment to 1.7 percent during the 19 to 24
month enrollment span having an inpatient visit. The regression analysis suggests that
there are conflicting impacts of Health Home participation on ED and inpatient
utilization, estimating that participation is related to a statistically significant increase
in ED visits and a decrease in inpatient visits between CY 2013 and CY 2015.
The descriptive analysis and regression analysis both suggest that participants
who received care from MTS providers had a higher percentage of inpatient
hospitalizations, ED visits, and 30-day all-cause hospital readmissions when
compared with those who received care from OTP or PRP providers. This may be
iii
due in part to the fact that the MTS population is higher risk than the other Health
Home participants.
The regression results suggest that those in the Health Home program have 50
percent higher annual health care costs than those in the comparison group at
baseline and that participating in the Health Home program is related to a 24
percent increase in total annual health care costs between CY 2013 and CY 2015,
holding all other variables constant.
1
2016 Health Home Evaluation Report
Introduction
Section 1 of this report provides background information on the Health Home program as a
whole, including an overview of the implementation of Health Homes in other states. Section 2
details the progress of the Maryland Health Home program, including descriptive statistics of
participant characteristics between Health Homes. Section 3 describes Health Home
participants’ patterns of health care utilization. Section 4 provides a comparison of outcomes
between Health Home participants and a comparison group comprised of similar Medicaid
participants.
Section 1. The Health Home Model
Background
Health Homes are intended to improve health outcomes for individuals with chronic conditions
by providing patients with an enhanced level of care management and care coordination. The
Affordable Care Act (ACA) created the option for state Medicaid programs to establish Health
Homes.3 Health Homes provide an integrated model of care that coordinates primary, acute,
behavioral health, and long-term services and supports for Medicaid participants who have two
or more chronic conditions, one chronic condition and a risk for developing a second chronic
condition, or a serious and persistent mental illness. In response to this initiative, DHMH
submitted a Medicaid SPA that was approved by the Centers for Medicare & Medicaid Services
(CMS) effective October 1, 2013.
The concept of the Health Home evolved from the Medical Home model, introduced by the
American Academy of Pediatrics in 1967 to provide more centralized care for children with
special health care needs. While a ―Medical Home‖ initially denoted a single source for all of a
patient’s medical information, it came to refer more broadly to an approach to primary care that
is comprehensive, coordinated, and patient- and family-centered (Sia, Tonninges, Osterhus, &
Taba, 2004). In 2007, four primary care specialty societies (the American Academy of
Physicians, the American Academy of Family Physicians, the American College of Physicians,
and the American Osteopathic Association) agreed on the Joint Principles of the Patient-
Centered Medical Home (PCMH) (Higgins, Chawla, Colombo, Snyder, & Nigam, 2013). The
PCMH was to include a personal physician, a physician-directed medical practice, a whole-
person orientation, coordination across providers and specialties, safe and high-quality care,
enhanced access to care, and payment that recognized the benefit provided to patients who have
a patient-centered medical home (American Academy of Family Physicians (AAFP), American
3 ACA § 2703(a) (42 USC § 1396w-4(a)).
2
Academy of Pediatrics (AAP), American College of Physicians (ACP), American Osteopathic
Association (AOA), 2007).
There has been growing recognition of the fragmentation between behavioral health and primary
care faced by individuals with mental health and/or SUDs, who are more likely to die
prematurely from untreated and preventable chronic illnesses (Scott, & Happell, 2011).
According to CMS, Medicaid is ―the single largest payer for mental health services in the United
States and is increasingly playing a larger role in the reimbursement of SUD services‖ (CMS,
2014). Additionally, Medicaid beneficiaries with serious mental illnesses (SMIs) and SUDs are
more likely to have co-occurring chronic conditions than are similar Medicaid beneficiaries
(Dickey, Normand, Weiss, Drake, & Azeni, 2002). These issues provide the motivation to
examine the impact of additional care coordination and care management services on the health
outcomes of vulnerable populations.
Health Home Programs Nationwide
As of November 2016, CMS approved 29 Health Home programs submitted by 20 states and the
District of Columbia between 2011 and 2016 (CMS, 2016). Enrollment in these programs varies
from less than 1,000 to over 500,000 participants. A majority of the programs are focused on
participants with an SMI and/or an SUD. A significant proportion of programs have a broad
focus, serving participants with chronic conditions. Two states have programs that are aimed at
children with a serious emotional disturbance (SED). One state targets participants with
HIV/AIDs. While some states have elected to auto-enroll (opt-out enrollment) all eligible
Medicaid participants into the Health Home, other states require participants to actively choose
to enroll (opt-in enrollment) and complete an intake process with a provider (CMS, 2016).
States are required to engage in activities to monitor the implementation and outcomes of their
Health Home model. CMS established a multi-pronged approach to evaluating Health Homes.
The data reporting requirements common to all states include a core set of eight metrics that
were selected by CMS (CMS, 2010). These metrics target chronic disease, behavioral health, and
appropriate utilization of health care. In order to implement a Health Home program, states
submit a two-year SPA to CMS, during which time they receive an enhanced federal medical
assistance percentage (FMAP) for the services provided. As part of their SPA, states outline their
methodology for monitoring quality improvement, health care utilization, and the cost of care
pertinent to their programs.
In addition to the reporting completed by the states, CMS is working with the U.S. Department
of Health and Human Services Office of the Assistant Secretary for Planning Evaluation
(ASPE) to conduct an independent evaluation of SPAs approved during the first three years of
the evaluation (Urban Institute, 2012). A five-year analytical plan is in place that began in
October 2011. The initial three years of the evaluation focuses on implementation, while the
fourth and fifth year will measure changes in quality, cost, utilization, and health outcomes of
3
program recipients compared with non-participants. The evaluation will be used to develop a
report to Congress in 2017 (Department of Health and Human Services, 2014).
In December 2016, DHMH reviewed Medicaid websites for states with an approved SPA to
locate interim reports describing the implementation and outcomes of their Health Homes. In
addition to Maryland, reports were published by Missouri, Maine, Iowa, Washington,
Minnesota, and West Virginia. There was a range of different evaluations presented, reflecting
the diversity of Health Home programs developed by these states. All states provided
descriptions of their participant populations, including demographics, clinical characteristics,
and enrollment data. The states selected various metrics to evaluate their programs but also
incorporated some of the core measures designated by CMS. The metrics selected included
Healthcare Effectiveness Data and Information Set (HEDIS)-derived outcome measures
focusing on monitoring chronic disease management, emergency department (ED) visits, and
total cost of care (Department of Mental Health and MO Healthnet, 2013; Momany, Damiano,
Bentler, McInroy, & Nguyen-Hoang, 2014).
Only three states—Missouri, Iowa, and Minnesota—had sufficient data available to offer post-
intervention information in their evaluation (Momany, Damiano, Bentler, McInroy, & Nguyen-
Hoang, 2014; Momany, Damiano, & Bentler, 2014; Department of Mental Health and MO
Healthnet, 2013; Momany, Damiano, & Bentler, 2014; Wholey, D. R., Finch, M., Shippee, N.
D., White, K. M., Christianson, J., Kreiger, R., et al., 2016). All three reports offered preliminary
results suggesting that their Health Home programs had effects on utilization and costs per
Medicaid participant. The authors noted mixed results, with improvements in certain areas (e.g.
reductions in ED visits and decreases in per member per month costs), but less or negative
impact in other areas (e.g. preventive care visits). Caution must be used when interpreting these
results. Each report applied different methods for conducting their analyses, used varying
approaches in how they selected participants to include in the study, and may not have had
sufficient time to detect changes in long-term health outcomes.
4
Section 2. The Maryland Health Home Program
The Maryland Health Homes program builds on statewide efforts to integrate somatic and
behavioral health services, with the aim of improving health outcomes and reducing avoidable
hospital utilization. The program targets populations with behavioral health needs who are at
high risk for additional chronic conditions, offering them enhanced care coordination and
support services from providers from whom they regularly receive care. The program is focused
on Medicaid participants with a serious and persistent mental illness (SPMI), an opioid SUD and
risk of additional chronic conditions due to tobacco, alcohol, or other non-opioid substance use,
and children with SED (CMS, 2013). In a Health Home, the center of a patient’s care, instead of
being in a somatic care setting, is in MTSs, PRPs, and OTPs. This service delivery method is
intended to include nurses and somatic care consultants into these programs and to make sure
individuals in MTS, PRPs, and OTPs receive improved somatic care.
Medicaid participants can enroll in Health Homes if they are eligible for and engaged with a
PRP, MTS, or an OTP that has been approved by DHMH to function as a Health Home provider.
Instead of auto-enrollment into the program, Maryland requires participants to actively choose to
enroll and complete an intake procedure. In order to improve care coordination when enrolling
into the Health Home, Medicaid participants are also required to consent to have their data
shared with the Chesapeake Regional Information System for our Patients (CRISP), a regional
health information exchange (HIE) serving Maryland and the District of Columbia. Individuals
are excluded from Health Home participation if they are currently receiving other Medicaid-
funded services that may duplicate those provided by Health Homes, such as targeted mental
health care management.
Health Home providers must be enrolled as a Maryland Medicaid provider and accredited as a
Health Home. A dedicated care manager must be assigned to each participant, and providers are
required to maintain certain staffing levels based on the number of participants. The Health
Home staff must include a Health Home director, physician, and nurse practitioner. Health
Homes are responsible for documenting all services delivered, participant outcomes, and social
indicators in the eMedicaid care management system. They must notify each participant’s other
providers of the participant's goals and the types of services the individual is receiving via the
Health Home and encourage participation in care coordination efforts.
Figures 1a and 1b display the number of participating Health Home providers and provider sites
by month. These data only include Health Home provider organizations that had at least one
participant enrolled during that month. A small number of providers were active at the inception
of the program. Within the first six months, the number of providers tripled. This number of
participating providers remained stable in the second half of 2014, increased by six providers in
2015, and increased slightly throughout 2016.
5
Figure 1a. Number of Participating Health Home Providers, by Month
Figure 1b displays the number of participating Health Home providers by month according to the
number of individual sites that are operational. These data only include Health Home sites that
had at least one participant enrolled during that month. A small number of providers were active
at the inception of the program—8 providers across 13 sites. Within the first six months the
number of Health Home provider sites tripled to 39. The number of participating sites continued
to increase in 2014 and through 2015. The number of Health Home provider sites has remained
relatively steady since October 2015, ranging between 66 and 67 provider sites between then and
June 2016.
Figure 1b. Number of Participating Health Home Provider Sites, by Month
6
Health Home Data
This report presents measures that were selected to provide an overview of the patient outcomes
for participants in Health Homes. The measures were calculated using data that Health Home
providers entered into the eMedicaid care management data system and data from the Maryland
Medicaid Information System (MMIS2).
eMedicaid is a secure web-based portal that allows health care practitioners to enroll as a
Medicaid provider, verify recipient eligibility, obtain payment information, and serve as a
care management tracking tool for providers participating in Maryland’s Health Home
program. Within eMedicaid, providers enroll participants and report participants’ diagnoses,
outcomes, and services rendered. The measures of participant characteristics and Health
Home services in the tables below are calculated from data reported by Health Home
providers into the eMedicaid care management system.
Participant Characteristics
Figure 2 presents enrollment data for the first ten quarters of the program. Enrollment is
determined using data reported by Health Home providers into the eMedicaid care management
system as of September 22, 2016. Figure 2 shows that a large portion of participants enrolled
near the start of the program. While the enrollment of new participants dropped after the months
immediately following implementation, new participants were continuously added every quarter,
resulting in enrollment more than doubling between Quarters 1 and 7. During the most recent
two years of the evaluation period, an average of more than 500 participants enrolled in the
program during each quarter. This increase in Health Home participant enrollment is primarily
due to the introduction of new provider sites, as the sizes of individual provider sites tend to
remain stable after an initial ramp-up period.
7
Figure 2. Number of Health Home Participants, by Enrollee Type and Quarter
Figure 3 presents enrollment data by program type: PRP, MTS, or OTP. PRP providers
consistently enrolled the largest share of Health Home participants – between 74 percent and 83
percent of participants each quarter across all 11 quarters. The percentage of participants enrolled
in the MTS program ranged between 3.5 percent and 6.6 percent across the intervention quarters,
while OTP enrollment ranged between 10.5 percent and 21.8 percent. As of Quarter 10, only
three of the 38 providers offer care to participants through multiple program types. The
remaining providers offer services as one program type.
8
Figure 3. Number of Health Home Participants, by Program Type and Quarter
Table 1 presents the percentage of Health Home participants by various demographic
characteristics. The largest proportion of participants was aged 40 to 64 years (56.9 percent),
followed by those aged 21 to 39 years (25.4 percent). Approximately 13 percent of the
participants were under the age of 21 years. Table 1 also shows that the vast majority of the
Health Home population identified as either White (41.1 percent) or Black (47.3 percent). Those
who identified as Other/Unknown, Asian, or Hispanic made up a small proportion (11.6 percent)
of total participants. A slight majority of Health Home participants were male (54.0 percent).
The region with the majority of participants was the Baltimore metropolitan area, with 64.8
percent of all Health Home participants. The next most common areas of residence were the
Eastern Shore (17.1 percent) and Montgomery and Prince George’s Counties (10.7 percent).
A person’s co-morbidity level is estimated based on the Johns Hopkins Adjusted Clinical Groups
(ACG) methodology, which uses claims data to classify individuals based on their projected
and/or actual utilization of health care services. Approximately 57.4 percent of participants were
categorized as having a very high or high co-morbidity level, 37.5 percent were classified as
having a moderate co-morbidity level, and only 5.1 percent were classified as having a low co-
morbidity level. Almost a third (32.1 percent) of Health Home participants were dually eligible
for both Medicare and Medicaid.
9
Table 1. Demographic and Clinical Characteristics of Health Home Participants Demographic/Clinical Characteristics Health Home Participants
Number Percentage
Age Group (Years) 3 to 9 208 2.8%
10 to 14 469 6.3%
15 to 20 285 3.8%
21 to 39 1,896 25.4%
40 to 64 4,251 56.9%
Race/Ethnicity
Asian 94 1.3%
Black 3,537 47.3%
White 3,075 41.1%
Hispanic 57 0.8%
Other/Unknown 710 9.5%
Gender
Female 3,437 46.0%
Male 4,035 54.0%
Transgender * 0.0%
Region
Baltimore Metro 4,844 64.8%
Eastern Shore 1,281 17.1%
Montgomery and Prince George's County 802 10.7%
Southern Maryland * 0.1%
Western Maryland 525 7.0%
Out of State * 0.1%
Adjusted Clinical Groups Co-Morbidity Level Low Co-morbidity 384 5.1%
Moderate Co-morbidity 2,804 37.5%
High Co-morbidity 2,045 27.4%
Very-High Co-morbidity 2,240 30.0%
Dual Medicaid-Medicare Eligibility
No 5,072 67.9%
Yes 2,401 32.1%
Total 7,473
Health Home Services
Health Homes are required to provide at least two services to a participant in a given month in
order to qualify for a $100.85 monthly rate per participant. Health home services include care
coordination, care management, health promotion, and referrals to community and social support
10
services. Categories of services include: (1) comprehensive care management to assess, plan,
monitor, and report on participant health care needs and outcomes; (2) care coordination to
ensure appropriate linkages, referrals, and appointment scheduling across different providers;
(3) health promotion to aid participants in implementation of their care plans; (4) comprehensive
transitional care to ease the transition when discharged from inpatient settings and ensure
appropriate follow-up; (5) individual and family support services to provide support and
information that is language, literacy, and culturally appropriate; and (6) referral to community
and social support services.
Figure 4 displays the percentage of participants by the number of services received per month.
During the first month of the program, 12.6 percent of participants received two or more services
and 75.2 percent of participants did not receive any services. As time progressed, the number of
participants receiving two or more services per month increased, ranging from 63.1 to 83.0
percent. A corresponding decrease in the number of participants who did not receive any services
is also noted. The percentage of participants not receiving any services ranged from 8.1 to 32.1
percent in each month between November 2013 and August 2016.
11
Figure 4. Percentage of Health Home Participants Receiving 0, 1, or 2 or More Services, by Month
173
165
496
610
690
557
403
561
499
572
452
471
435
499
360
384
510
405
372
327
322
350
357
404
481
532
497
495
509
565
670
644
846
846
1,473
2831
74
135
152
360
327
281
233
345
288
237
210
281
332
251
277
329
320
410
350
466
449
581
426
453
382
394346
269
278312
225
262
274
29
580
974
1,343
1,5041,728
2,119
2,2322,450
2,395
2,6962,789
2,889
2,8262,886
2,955
2,835
2,891
3,093
3,117
3,2823,226
3,334
3,267
3,435
3,442
3,570
3,658
3,791
3,8933,867
3,946
3,916
3,9443,211
75.2%
21.3%
32.1%
29.2%
29.4%
21.1%
14.1%
18.2%
15.7%
17.3%
13.2%
13.5%
12.3%
13.8%
10.1%
10.7%
14.1%
11.2%
9.8%
8.5%
8.1%
8.7%
8.6%
9.5%
11.1%
12.0%
11.2%
10.9%
11.0%
12.0%
13.9%
13.1%
17.0%
16.7%
29.7%
12.2%
4.0%
4.8%
6.5%
6.5%
13.6%
11.5%
9.1%7.3%
10.4%
8.4%6.8%
5.9%
7.8%
9.3%7.0%
7.6%
9.1%
8.5%
10.6%
8.9%
11.5%
10.8%
13.7%
9.8%
10.2%
8.6%
8.7%
7.4%
5.7%
5.8%6.4%
4.5%5.2%
5.5%
12.6%
74.7%
63.1%
64.3%
64.1%
65.3%
74.4%
72.6%
77.0%
72.3%
78.5%
79.8%
81.7%
78.4%
80.7%
82.3%
78.3%
79.8%
81.7%
80.9%
83.0%
79.8%
80.5%
76.8%
79.1%
77.8%
80.2%
80.4%
81.6%
82.4%
80.3%
80.5%
78.5%
78.1%
64.8%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
October 2013
November 2013
December 2013
January 2014
February 2014
March 2014
April 2014
May 2014
June 2014
July 2014
August 2014
September 2014
October 2014
November 2014
December 2014
January 2015
February 2015
March 2015
April 2015
May 2015
June 2015
July 2015
August 2015
September 2015
October 2015
November 2015
December 2015
January 2016
February 2016
March 2016
April 2016
May 2016
June 2016
July 2016
August 2016
Percentage of Participants
Tim
e (M
on
ths)
0 Services 1 Service 2+ Services
12
Figure 5 presents the average number of services among Health Home participants who received
at least one service during the quarter. The average number of services increased as the program
progressed, ranging from 3.0 in Quarter 1 to 6.3 in Quarter 8. The average number of services
decreased slightly after Quarter 8 to 5.5 in Quarter 11.
Figure 5. Average Number of Services Received by Health Home Participants, by Quarter
Figure 6 displays the percentage of participants who received at least one type of Health Home
service required by CMS. The figure demonstrates that there is a strong demand from
participants for the Health Home social services. Care coordination was consistently received at
least once per quarter by approximately half of the participants. The proportion of participants
receiving a comprehensive care management service increased from 33.6 percent in Quarter 1 to
80.6 percent in Quarter 4. The average proportion of participants receiving a comprehensive care
management service from Quarter 4 onwards remained at approximately 80 percent for the
following two years. Receipt of health promotion services increased from 36.9 percent in Quarter
1 to between 60.0 and 66.3 percent through the subsequent ten quarters. Comprehensive
transitional care and referral to community and social support services were consistently received
by the smallest proportion of participants.
13
Figure 6. Percentage of Health Home Participants by Types of Health Home Services Received, by Quarter
Table 2 presents the distribution of Health Home participants’ first and last clinical body mass
index (BMI) recordings, categorized by BMI groupings aligned with HEDIS 2016 reporting
thresholds.4 The cohort is comprised of Health Home participants with at least two clinical BMI
entries during their enrollment span. The greatest percentage of both first and last BMI
recordings fell within the 25.0-29.9 BMI range – 26.7 percent of the first BMI recordings and
26.5 percent of the last BMI recordings fell within this category. Approximately 5 percent of first
and last BMIs were recorded as less than 20.0. The largest difference between the first and last
recorded BMI was in the 20.0-24.9 category. Eighteen percent of enrollees’ first recorded BMI
were within 20-24.9, while in the last BMI recorded, 16.8 percent of enrollees fell within that
range.
4 Tables 2 and 3 present clinical results reported into the eMedicaid reporting system. These data have not been
modified to exclude possible outliers. Therefore, data entry errors may potentially skew the estimates.
14
Table 2. Number and Percentage of Health Home Participants with More than One Body Mass Index (BMI) Recorded, According to HEDIS BMI Thresholds
First BMI Last BMI
BMI Grouping Number Percent Number Percent
<20 201 5.1% 214 5.5%
20-24.9 705 18.0% 656 16.8%
25-29.9 1,045 26.7% 1,035 26.5%
30-34.9 865 22.1% 885 22.6%
35-39.9 524 13.4% 545 13.9%
≥40 573 14.6% 578 14.8%
Total 3,913 100.0% 3,913 100.0%
Table 3 presents the distribution of Health Home participants’ first and last clinical blood
pressure (BP) recordings, categorized by systolic and diastolic BP groupings that are aligned
with HEDIS 2016 reporting thresholds. The cohort is comprised of Health Home participants
that have had at least two BP values recorded in eMedicaid during their enrollment span. For
systolic BP entries, nearly 80 percent of both the first and last BP entries were less than 140 and
20 percent were greater than or equal to 140. In both their first and last entries, nearly 50 percent
of the diastolic BP entries recorded were less than 80, approximately 34 percent fell within the
range of 80 to 89, and approximately 17 percent were greater than or equal to 90.
Table 3. Number and Percentage of Health Home Participants with More than One Blood Pressure (BP) Recorded, According to HEDIS BP Thresholds
First BP Last BP
Systolic BP Grouping
Number of Participants
Percentage of Participants
Number of Participants
Percentage of Participants
<140 3,057 79.9% 3,051 79.7%
≥140 769 20.1% 775 20.3%
Total 3,826 100.0% 3,826 100.0%
Diastolic BP Grouping
Number of Participants
Percentage of Participants
Number of Participants
Percentage of Participants
<80 1,866 48.8% 1,894 49.5%
80-89 1,295 33.8% 1,300 34.0%
≥90 665 17.4% 632 16.5%
Total 3,826 100.0% 3,826 100.0%
15
Section 3. Health Home Participants Health Care Utilization Patterns
In contrast with the service data presented above, the values reported in the following sections
are based on claims and encounters reported in the MMIS2. The figures and tables below
describe the overall composition of health care services received by Health Home participants
and categories of types of ED visits during calendar year (CY) 2013, CY 2014, and CY 2015.
Those are followed by tables describing Health Home participants’ inpatient and ED visit rates
according to participants’ length of enrollment in the Health Home program.
Composition of Total Medicaid Services
Figure 7 presents the overall composition of Medicaid services received by Health Home
participants. The services were grouped into the following categories: prescriptions, behavioral
health services, and all other services. In CY 2013, behavioral health services accounted for 42.1
percent of all Medicaid services received by Health Home participants. The proportion of
behavioral health services increased to 47.5 percent in CY 2014 and then declined to 45.2
percent in CY 2015. The ―all other services‖ category remained about the same across the study
period at slightly less than 30 percent. Prescriptions dropped from 28.3 percent in CY 2013 to
22.8 percent in CY 2014, then increased to 24.8 percent in CY 2015.
Figure 7. Composition of Types of Services Received by Health Home Participants, by Calendar Year
16
Emergency Department Utilization Patterns
Table 4 presents the percentage of Health Home participants who had at least one ED visit in CY
2013 though CY 2015. The table also shows the percentages of participants who visited the ED
for behavioral health services and/or somatic care. To identify those ED visits related to
behavioral health, the team used a grouping method based on classifications developed by the
New York University (NYU) Center for Health and Public Service Research (Billings, Parikh, &
Mijanovich, 2000).
In CY 2013, 34.5 percent of ED users had an ED visit with a diagnosis related to behavioral
health; this percentage increased to 36.9 percent in CY 2014 and then decreased to 34.4 percent
in CY 2015. A greater percentage of ED users, 89.2 percent, visited the ED for somatic care in
CY 2015, compared to 88.7 percent in CY 2014 and 89.0 percent in CY 2013. Please note that
participants in the table could have seen more than one type of provider . These categories are
not mutually exclusive; therefore, the sum of the frequencies does not equal the total number of
participants with any ED visit.
Table 4. Percentage of Health Home Participants Completing ED Visits, by Service Type,
CY 2013 – CY 2015 Year Total
Participants Number of
Participants with Any ED
Visit
Percentage with Any ED
Visit
Number of Participants
with a Behavioral Health ED
Visit
Percentage of ED Users
with a Behavioral Health ED
Visit
Number of Participants
with a Somatic Care ED
Visit
Percentage of ED Users
with a Somatic Care ED
Visit
CY 2013 7,473 3,679 49.2% 1,271 34.5% 3,274 89.0%
CY 2014 7,473 4,198 56.2% 1,551 36.9% 3,723 88.7%
CY 2015 7,473 4,090 54.7% 1,408 34.4% 3,647 89.2%
Figure 8 displays the distribution of Health Home participants’ total number of ED visits,
categorized by the type of care provided. Of the total ED visits in CY 2013, 24.0 percent were
for behavioral health and 76.0 percent were for somatic care. A similar trend occurred in CY
2014. In CY 2015, behavioral health ED visits increased to 25.6 percent of all ED visits, while
somatic care ED visits made up 74.4 percent.5
5 A preliminary analysis conducted by the Maryland Behavioral Health Network (MBHN) on a subset of Health
Home providers suggests that there may have been decreases in non-behavioral health related health care service
utilization rates, when comparing the period of January 1 through March 31 of CYs 2014 and 2015. These results
are preliminary, and the methodology used to generate these rates has not been verified.
17
Figure 8. Number and Composition of ED Visits Received by Health Home Participants, CY 2013 – CY 2015
Health Care Utilization by Length of Enrollment
Tables 5 and 6 present ED and inpatient utilization rates per participant by length of enrollment
in a Health Home program during CY 2013 through CY 2015. The tables below summarize
health care utilization patterns while participants were enrolled in the Health Home program.6
The lengths of enrollment were calculated as of the end of CY 2015 and combine the time
periods of participants with more than one enrollment span. As of December 31, 2015, the
average length of enrollment in the Health Home program was 14 months.
ED utilization rates were highest during a participant’s first six months of enrollment, with 37.6
percent of total participants visiting the ED at least one time during that enrollment span. The ED
utilization rate declined the longer those participants stayed in the Health Home program.
Participants who were in a Health Home program between 19 to 24 months had the lowest ED
utilization rate at 27.5 percent of participants with at least one ED visit during that enrollment
span. Furthermore, the average number of ED visits per participant decreased the longer
6 If a participant were discharged from the Health Home program, later visited the ED, and subsequently re-enrolled
in the program, that visit is not included in the tables below
14,096 16,770 16,074
3,380 4,059 4,109
10,716 12,711 11,965
24.0% 24.2% 25.6%
76.0% 75.8% 74.4%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
CY 2013 CY 2014 CY 2015
Time (CY)
Pe
rce
nta
ge o
f ED
Vis
its
% Behavioral Health ED Visits % Somatic Care ED Visits
18
participants were enrolled in the program from 0.96 during the first six months of enrollment, to
0.68 when participants were enrolled 19 to 24 months.
Table 5. ED Utilization Rates per Participant, by Length of Enrollment Length of
Enrollment Total
Participants Number with Any ED Visit
Percent ED Utilization
Number of ED Visits
Average ED Visits per Participant
0 to 6 Months 6,613 2,487 37.6% 6,348 0.96
7 to 12 Months 5,033 1,656 32.9% 4,043 0.80
13 to 18 Months 3,633 1,126 31.0% 2,808 0.77
19 to 24 Months 2,617 720 27.5% 1,767 0.68
Table 6 presents the inpatient utilization rates per participant by length of enrollment. Inpatient
utilization rates were highest during participants’ first six months in the program, with 10.0
percent of total participants visiting the hospital during that enrollment span. The inpatient
utilization rate declined the longer participants were enrolled in the Health Home program.
Participants who remained in a Health Home program for 19 to 24 months had the lowest
inpatient utilization rate at 1.7 percent of total participants with any inpatient visit.
Table 6. Inpatient Utilization Rates per Participant, by Length of Enrollment
Length of Enrollment
Total Participants
Number with Any Inpatient Visit
Percent Inpatient
Utilization
0 to 6 Months 6,613 660 10.0%
7 to 12 Months 5,033 486 9.7%
13 to 18 Months 3,633 302 8.3%
19 to 24 Months 2,617 45 1.7%
19
Section 4. Health Home Participant Outcomes
This section of the report presents a comparison of the health care utilization, quality, and costs
between Health Home participants and a comparison group of Medicaid participants. The
analysis began with a description of the selection of the participant and comparison groups,
followed by an overview of the groups’ demographic characteristics, and then a presentation of
the utilization, quality, and cost outcomes of interest.
Evaluation Cohort Description
The team selected a sub-population of the Health Home and other Medicaid participants to use as
study and comparison groups for this evaluation in order to help estimate the effects of the
program. Estimating the same measures between carefully selected groups of similar people can
aid in distinguishing changes associated with participation in the Health Home program from
changes due to other contributing factors. To identify the comparison group of interest, the team
first created a sub-group of Health Home and other Medicaid participants that met the following
criteria:
1. Aged 18 to 64 years
2. Were continuously enrolled in Medicaid from CY 2013 through CY 2015
3. Received care in CY 2012 from a provider of the same type as a Health Home
provider, in order to estimate the outcomes of participants with similar health needs.
These provider types include:
a. Drug Clinics (Provider Type 32)
b. Mobile Treatment Programs (Provider Type MT)
c. Psychiatric Rehab Services Facilities (Provider Type PR)
Once the selection of potential comparison group members was completed, the team used a
propensity score matching statistical technique to select an evaluation cohort, i.e., a study and
comparison group, in which the likelihood of joining the program is as similar as possible
between the two groups. The likelihood of joining the program was estimated based on a
participant’s geographic region of residence, age, race/ethnicity, gender, ACG co-morbidity
grouping, and type of Health Home provider seen. A detailed description of the process used to
select the evaluation cohort is presented in Appendix A.
20
Demographic Characteristics
Table 7 provides an assessment of the study and comparison groups of several demographic and
clinical characteristics. Overall, the propensity score matching technique produced a comparison
group that was similar to the study group. In both groups, a majority of the participants were
between the ages of 40 and 64 years, resided in the Baltimore metropolitan area, and were
categorized as having a moderate to very-high co-morbidity level. Those in the study group are
slightly younger and less likely to be an Other/Unknown race than those in the comparison
group.
The characteristics of the study group differ from the wider Health Home population described
earlier in the report. People in the study group are older, more likely to be Black, more likely to
be female, less likely to be from the Eastern Shore, less likely to have low co-morbidities, and
are more likely to be dually eligible for Medicare and Medicaid than people in the overall Health
Home program.
21
Table 7. Demographic and Clinical Characteristics of Study Group and Comparison Group Descriptive
Characteristics Health Home Study Group Comparison Group
Number Percentage Number Percentage
Age Group (Years)
Under 21 61 1.9% 71 2.2%
Ages 21 to 39 961 29.2% 827 25.1%
Ages 40 to 64 2,268 68.9% 2,392 72.7%
Race/Ethnicity
Asian 57 1.7% 31 0.9%
Black 1,612 49.0% 1,571 47.8%
White 1,401 42.6% 1,410 42.9%
Hispanic 31 0.9% 25 0.8%
Other/Unknown 189 5.7% 253 7.7%
Gender
Female 1,618 49.2% 1,593 48.4%
Male 1,672 50.8% 1,697 51.6%
Region
Baltimore Metro 2,117 64.4% 2,162 65.7%
Eastern Shore 342 10.4% 318 9.7%
Montgomery and Prince George's County
519 15.8% 505 15.4%
Southern Maryland * * * *
Western Maryland 308 9.4% 294 8.9%
Out of State * * * *
Adjusted Clinical Groups Co-Morbidity Level
Low Co-morbidity 28 0.9% 86 2.6%
Moderate Co-morbidity 1,364 41.5% 1,406 42.7%
High Co-morbidity 951 28.9% 922 28.0%
Very-High Co-morbidity 947 28.8% 876 26.6%
Dually Eligible
No 2,106 64.0% 2,164 65.8%
Yes 1,184 36.0% 1,126 34.2%
Total 3,290 3,290
Table 8 compares the distribution of the study group and the comparison group by program type
of their respective providers. PRP providers were seen by the largest proportion of both the study
and comparison groups, at 77.4 and 83.7 percent of participants, respectively. Please note that
the people in the comparison group could have seen more than one type of provider; these
categories are not mutually exclusive. Therefore, the sum of the frequencies does not equal the
total comparison group population.
22
Table 8. Program Types of the Study Group and Comparison Group
Provider Type Health Home Study Group Comparison Group
Number Percentage Number Percentage
Psychiatric Rehabilitation Facility 2,545 77.4% 2,754 83.7%
Mobile Treatment Services 127 3.9% 157 4.8%
Opioid Treatment Program 618 18.8% 426 13.0%
Total 3,290 3,290
Health Care Utilization Outcomes
All measures presented in the tables below include the study and comparison groups detailed
above. In order to generate comparable results, all percentages have been weighted to account for
the matching technique and sample size difference between the study and comparison groups. A
description of the analytical methods used is included in Appendix A.
Inpatient Hospital Admissions
Table 9 compares the percentage of Health Home study and comparison group participants with
at least one inpatient hospital admission in CY 2013, CY 2014, and CY 2015. In CY 2013, 24.5
percent of people in the study group had at least one inpatient hospital admission; this increased
slightly to 25.5 percent in CY 2014 and then decreased to 24.4 percent in CY 2015. Within the
study group, the percentage of participants who were dually eligible for Medicaid and Medicare
and had at least one inpatient hospital admission decreased from 23.9 percent in CY 2013 to 21.9
percent in CY 2015. The percentage of those who were not dually eligible for Medicare and
Medicaid and had at least one inpatient hospital admission increased from 24.8 percent in CY
2013 to 25.9 percent in CY 2015.
In CY 2013, 21.2 percent of people in the comparison group had at least one inpatient hospital
admission; this increased to nearly 23 percent in both CY 2014 and CY 2015. The percentage of
participants in the comparison group who were dually eligible for Medicaid and Medicare and
had at least one inpatient hospital admission remained stable from CY 2013 to CY 2015 at 21
percent. The percentage of participants in the comparison group who were not dually eligible and
had an inpatient admission increased from 21.0 percent in CY 2013 to 23.6 percent in CY 2015.
Throughout the evaluation period, the group of participants receiving services from MTS
providers tended to have higher rates of inpatient hospital admissions, but this group experienced
the largest decrease in inpatient utilization from CY 2013 to CY 2015 compared to those
enrolled in PRPs and OTPs.
23
Table 9. Percentage of Participants with at Least One Inpatient Hospital Admission, by Treatment Group, CY 2013 – CY 2015
Inpatient Hospital Admissions
Health Home Study Group Comparison Group
Provider
Type CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP 21.9% 27.9% 28.7% 19.2% 24.5% 23.3%
MTS 42.3% 33.7% 27.7% 35.2% 25.3% 34.3%
PRP 24.8% 25.5% 24.5% 20.7% 23.9% 23.1%
Total 24.8% 26.6% 25.9% 21.0% 24.0% 23.6%
Duals OTP 27.0% 26.3% 25.0% 33.3% 37.2% 36.4%
MTS 43.5% 26.9% 30.8% 34.6% 20.7% 27.6%
PRP 23.4% 22.2% 21.6% 20.5% 20.5% 20.3%
Total 23.9% 23.6% 21.9% 21.6% 21.1% 21.2%
All Participants
OTP 22.2% 27.8% 28.5% 20.4% 25.8% 24.7%
MTS 42.5% 32.3% 28.4% 35.0% 23.6% 31.9%
PRP 24.2% 24.6% 23.1% 20.6% 22.6% 22.0%
Grand Total 24.5% 25.5% 24.4% 21.2% 22.9% 22.7%
Lengths of Inpatient Hospital Stay
Table 10 compares the average length of stay (in days) in CY 2013 though CY 2015 for the
Health Home study and comparison groups. For the Health Home study group, the average
length of stay was nearly the same for all three years (13.0 days in CY 2013 and CY 2014, and
13.1 days in CY 2015). Within the study group, those who were dually eligible for Medicaid
and Medicare had slightly lower average lengths of stay than those who were not dually
eligible. In CY 2013, the average length of stay for dually-eligible participants was 11.8 days
and 13.6 days for non-duals; in CY 2014, the average length of stay for dually-eligible
participants was 12.7 days and 13.2 days for non-duals; and, in CY 2015, the average length of
stay for dually-eligible participants was 12.4 days and 13.5 days for non-duals. While the
length of stay fell for those in the MTS study groups between CY 2013 and CY 2015, the
average length of stay rose for participants in the OTP and PRP study groups during the
period.
The average length of stay for the comparison group varied slightly over the years. For the
comparison group as a whole, the average length of stay was 12.0 days in CY 2013, 13.0 days in
CY 2014, and 12.3 days in CY 2015. A similar trend was seen in those who were dually-eligible
for Medicare and Medicaid. Dualeligibles’ average length of stay was 11.1 days in CY 2013,
13.4 days in CY 2014, and 10.6 days in CY 2015. Non-duals experienced an increase in average
length of inpatient hospital stays from CY 2013 to CY 2015. Non-duals’ average length of stay
was 12.4 days in CY 2013, 12.9 days in CY 2014, and 13.2 days in CY 2015.
24
Table 10. Average Length of Stay for Inpatient Hospital Admission, by Treatment Group, CY 2013 – CY 2015
Inpatient Hospital Admissions
Health Home Study Group Comparison Group
CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Provider
Type
Number with One or More
Visits
Average Length of Stay (Days)
Number with One or More
Visits
Average Length of Stay (Days)
Number with One or More
Visits
Average Length of Stay (Days)
Number with One or More
Visits
Average Length of Stay (Days)
Number with One or More
Visits
Average Length of Stay (Days)
Number with One or More
Visits
Average Length of Stay (Days)
Non-Duals OTP 127 10.4 162 11.9 166 11.8 75 7.8 94 10.6 89 8.5
MTS 44 22.8 34 18.1 28 15.4 37 16.9 25 19.4 34 17.2
PRP 352 13.5 353 13.3 331 14.2 354 13.2 398 13.1 380 14.0
Total
523 13.6 549 13.2 525 13.5 454 12.4 506 12.9 493 13.2
Duals
OTP * 4.6 * 8.3 * 8.2 12 10.8 16 13.5 16 16.2
MTS * 17.1 * 23.3 * 17.5 18 10.2 12 22.3 16 6.9
PRP 263 11.9 272 12.6 258 12.4 214 11.3 223 12.8 225 10.5
Total
283 11.8 289 12.7 276 12.4 243 11.1 248 13.4 254 10.6
All Participants
OTP 137 10.0 172 11.7 176 11.6 87 8.2 110 11.0 105 9.7
MTS 54 21.7 41 19.0 36 15.8 55 14.7 37 20.3 50 13.9
PRP 615 12.9 625 13.0 589 13.4 568 12.5 621 13.0 605 12.7
Grand Total 806 13.0 838 13.0 801 13.1 697 12.0 754 13.0 747 12.3
25
Emergency Department Utilization
Table 11 displays the percentage of Health Home study and comparison group participants with
at least one ED visit. In CY 2013, 54.8 percent of study group participants had at least one ED
visit, rising to 56.6 percent in CY 2014 and then falling slightly in CY 2015 to 56.4 percent.
Among participants in the Health Home study group and comparison group receiving services
from an MTS provider, the percentage of those with an ED visit decreased by over 7 percentage
points between CY 2013 and CY 2015. Dually-eligible participants had lower rates of ED visits
than non-dual participants in all calendar years.
Table 11. Percentage of Participants with at Least One Emergency Department Visit, by Treatment Group, CY 2013 – CY 2015
ED Visits
Health Home Study Group Comparison Group
Provider
Type CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP 63.5% 70.2% 71.1% 63.3% 64.0% 62.0%
MTS 66.3% 59.4% 58.4% 66.7% 60.6% 64.6%
PRP 56.6% 58.4% 57.9% 56.6% 59.2% 55.3%
Total 59.0% 61.8% 61.7% 57.9% 59.8% 56.8%
Duals OTP 67.6% 68.4% 72.5% 63.9% 76.7% 65.9%
MTS 60.9% 65.4% 57.7% 67.3% 55.2% 50.0%
PRP 46.4% 46.9% 46.8% 46.0% 45.7% 43.3%
Total 47.4% 48.0% 47.8% 47.3% 47.0% 44.7%
All Participants
OTP 63.8% 70.1% 71.2% 63.4% 65.3% 62.4%
MTS 65.4% 60.6% 58.3% 66.9% 58.6% 59.2%
PRP 52.1% 53.2% 52.7% 52.5% 53.8% 50.6%
Grand Total 54.8% 56.6% 56.4% 54.3% 55.2% 52.3%
Ambulatory Care Utilization
An ambulatory care visit is defined as contact with a doctor or nurse practitioner in a clinic,
physician’s office, or hospital outpatient department.7 Ambulatory care utilization often
serves as a measure of access to care. Higher rates of ambulatory care can offer an alternative
to less efficient care for non-emergent conditions in an ED visit setting as well as prevent a
condition from becoming exacerbated to the extent that it requires an inpatient admission.
7 This definition excludes ED visits, hospital inpatient services, substance abuse treatment, mental health, home
health, x-ray, and laboratory services.
26
Table 12 presents the percentage of Health Home study and comparison group participants with
at least one ambulatory care visit. Over the evaluation period, the percentage of Health Home
participants in the study group with an ambulatory care visit was 94.0 percent in CY 2013, 95.2
percent in CY 2014, and 94.9 percent in CY 2015. The ambulatory care visit rate for the
comparison group dropped from 92.4 percent in CY 2013 to 90.5 percent in CY 2015.
In each calendar year, participants in both the study and comparison groups who received
services from a PRP provider had a higher rate of ambulatory care utilization compared to
participants receiving services from the other provider types. Dually-eligible participants in the
study group had higher rates of ambulatory care utilization than non-dual participants did in CY
2013 and CY 2014, but a slightly lower rate in CY 2015. Dually-eligible participants in the study
group had ambulatory care visit rates that were 5.0 percentage points higher than the control
group in CY 2013 and 6.3 percentage points higher in CY 2015.
Table 12. Percentage of Participants with at Least One Ambulatory Care Visit, by Treatment Group, CY 2013 – CY 2015
Ambulatory Care Visits
Health Home Study Group Comparison Group
Provider
Type CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP 85.0% 91.6% 91.2% 87.7% 87.7% 89.0%
MTS 83.7% 87.1% 88.1% 81.0% 87.9% 85.9%
PRP 97.2% 97.3% 97.3% 95.4% 94.4% 92.7%
Total 93.2% 95.2% 95.1% 93.3% 93.0% 91.7%
Duals OTP 78.4% 81.6% 87.5% 88.9% 83.7% 81.8%
MTS 91.3% 88.5% 76.9% 76.9% 75.9% 84.5%
PRP 96.3% 95.9% 95.2% 91.3% 89.7% 88.8%
Total 95.6% 95.3% 94.6% 90.6% 88.9% 88.3%
All Participants
OTP 84.6% 90.9% 90.9% 87.8% 87.3% 88.3%
MTS 85.0% 87.4% 85.8% 79.6% 83.4% 85.4%
PRP 96.8% 96.7% 96.3% 93.8% 92.5% 91.1%
Grand Total 94.0% 95.2% 94.9% 92.4% 91.5% 90.5%
Nursing Home Admissions
Table 13 presents information on nursing home stays. In CY 2013, 1.2 percent of participants in
the study group and 2.2 percent of participants in the comparison group had at least one nursing
home admission. In CY 2015, 2.4 percent of study group participants and 2.7 percent of
comparison group participants had a nursing home admission. OTP study group participants had
the largest increase in nursing home admissions, from 0.6 percent in CY 2013 to 2.9 percent in
CY 2015. In both the study and comparison groups, dually-eligible participants had, on average,
lower rates of nursing home admissions than non-dual participants did, with the exception of the
27
comparison group in CY 2015. For study group participants, the rate of dually-eligible
participants with at least one nursing home admission increased from 0.5 percent in CY 2013 to
1.5 percent in CY 2015. These rates were lower than the comparison group of dually-eligible
enrollees, which increased from 1.8 percent in CY 2013 to 2.9 percent in CY 2015.
Table 13. Percentage of Participants with at Least One Nursing Home Admission, by Treatment Group, CY2013 - CY2015
Nursing Home Admissions
Health Home Study Group Comparison Group
Provider
Type CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP 0.7% 3.3% 3.1% 2.1% 2.3% 1.6%
MTS 2.9% 2.0% 3.0% 2.9% 2.0% 1.0%
PRP 1.8% 2.0% 2.9% 2.6% 2.9% 3.0%
Total 1.6% 2.3% 3.0% 2.4% 2.7% 2.6%
Duals OTP 0.0% 0.0% 0.0% 2.8% 2.3% 4.5%
MTS 0.0% 0.0% 0.0% 1.9% 0.0% 3.4%
PRP 0.5% 0.9% 1.6% 1.7% 2.4% 2.9%
Total 0.5% 0.9% 1.5% 1.8% 2.3% 2.9%
All Participants
OTP 0.6% 3.1% 2.9% 2.1% 2.3% 1.9%
MTS 2.4% 1.6% 2.4% 2.5% 1.3% 1.9%
PRP 1.3% 1.5% 2.3% 2.3% 2.7% 2.9%
Grand Total 1.2% 1.8% 2.4% 2.2% 2.6% 2.7%
Health Care Quality Outcomes
30-Day All-Cause Hospital Readmissions
The 30-day all-cause hospital readmission rate, based on National Committee for Quality
Assurance (NCQA) definitions, was calculated as the percentage of acute inpatient stays during
the measurement year that were followed by an acute inpatient readmission for any diagnosis
within 30 days. The HEDIS 2013 specifications identify inclusion criteria for types of stays and
hospitals. The HEDIS specifications also limit the population to people continuously enrolled in
Medicaid with respect to the date of discharge.
Table 14 displays the percentage of Health Home study and comparison group participants with
at least one 30-day all-cause hospital readmission. Participants receiving services from MTS
providers had a greater likelihood of having a 30-day all-cause hospital readmission, occurring at
almost twice the rate in most years as those enrolled in OTP and PRP programs. Dually-eligible
28
participants had lower rates of 30-day all-cause hospital readmissions than non-dual participants
did in both the study and comparison groups across the study period.
Table 14. Percentage of Participant with at Least One 30-Day All-Cause-Hospital Readmission, by Treatment Group, CY2013 - CY2015
30-Day All-Cause Hospital Readmissions
Health Home Study Group Comparison Group
Provider
Type CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP 3.6% 6.7% 6.4% 3.8% 4.7% 2.1%
MTS 18.3% 11.9% 6.9% 12.4% 10.1% 12.1%
PRP 6.4% 7.3% 6.9% 6.1% 6.0% 6.3%
Total 6.2% 7.4% 6.8% 6.0% 5.9% 5.6%
Duals OTP 0.0% 2.6% 0.0% 5.6% 4.7% 6.8%
MTS 0.0% 7.7% 0.0% 1.9% 1.7% 0.0%
PRP 0.4% 0.6% 0.7% 0.8% 0.8% 0.2%
Total 0.4% 0.8% 0.6% 0.9% 1.0% 0.4%
All Participants
OTP 3.4% 6.5% 6.0% 4.0% 4.7% 2.6%
MTS 15.0% 11.0% 5.5% 8.9% 7.0% 7.6%
PRP 3.8% 4.2% 4.0% 4.1% 4.0% 3.8%
Grand Total 4.1% 4.9% 4.4% 4.2% 4.2% 3.7%
Appropriateness of ED Care
One widely used methodology to evaluate the appropriateness of care in the ED setting is based
on classifications developed by the New York University (NYU) Center for Health and Public
Service Research (Billings, Parikh, & Mijanovich, 2000). The algorithm assigns probabilities of
likelihoods that the ED visit falls into one of the following categories:
1. Non-emergent: Immediate care was not required within 12 hours based on patient’s
presenting symptoms, medical history, and vital signs.
2. Emergent but primary care treatable: Treatment was required within 12 hours, but it
could have been provided effectively in a primary care setting (e.g., CAT scan or certain
lab tests).
3. Emergent but preventable/avoidable: Emergency care was required, but the condition
was potentially preventable/avoidable if timely and effective ambulatory care had been
received during the episode of illness (e.g., asthma flare-up).
4. Emergent, ED care needed, not preventable/avoidable: Ambulatory care could not have
prevented the condition (e.g., trauma or appendicitis).
29
5. Injury: Injury was the principal diagnosis.
6. Alcohol-related: The principal diagnosis was related to alcohol.
7. Drug-related: The principal diagnosis was related to drugs.
8. Mental-health related: The principal diagnosis was related to mental health.
9. Unclassified: The condition was not classified in one of the above categories.
Table 15 presents the distribution of ―non-emergent‖ ED visits for the Health Home study and
comparison groups according to the NYU classification. If a visit is classified as more than 50
percent likely to fall into Categories 1 or 2 as described above, then it is considered ―non-
emergent.‖ The estimates presented in the table therefore show the percentage of participants
who went to the ED when either immediate care was not required within 12 hours or it could
have been provided in a primary care setting.
In CY 2013, 34.2 percent of the study group had a non-emergent ED visit, compared to 33.3
percent of the comparison group. The non-emergent ED visit rate increased for both groups in
CY 2014 and then decreased in CY 2015, to 33.4 percent for the study group and 32.0 percent
for the comparison group. During the evaluation period, participants receiving services from
OTP providers generally had higher rates of non-emergent ED visits than those receiving
services from the other provider types.
Table 15. Percentage of Participants with at Least One Non-Emergent ED Visit, by Treatment Group, CY2013 - CY2015
Non-Emergent ED Visits
Health Home Study Group Comparison Group
Provider
Type CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP 45.4% 49.8% 48.4% 39.5% 43.9% 39.5%
MTS 41.3% 37.6% 41.6% 41.0% 41.4% 42.4%
PRP 34.1% 35.4% 34.2% 36.4% 39.3% 34.4%
Total 37.6% 39.6% 38.6% 36.9% 40.2% 35.6%
Duals OTP 51.4% 57.9% 35.0% 41.7% 55.8% 40.9%
MTS 39.1% 30.8% 38.5% 34.6% 39.7% 17.2%
PRP 27.2% 26.4% 24.4% 25.7% 26.4% 25.4%
Total 28.2% 27.5% 25.1% 26.6% 27.8% 25.6%
All Participants
OTP 45.8% 50.3% 47.6% 39.7% 45.1% 39.7%
MTS 40.9% 36.2% 40.9% 38.9% 40.8% 33.1%
PRP 31.1% 31.3% 29.6% 32.3% 34.2% 30.8%
Grand Total 34.2% 35.0% 33.4% 33.3% 35.7% 32.0%
30
Health Care Cost Outcomes8
The following tables present preliminary data on the costs of health care for participants in the
study and comparison groups. The estimated costs for CY 2015 should be considered
preliminary due to the limited amount of time that has passed for adjudication of these claims
and encounters. The interim administrative claims and encounter data that are available at this
point may be missing payment information, and/or reversals of payment denials may not be
reflected. In addition, these data have not been revised to exclude outliers with extremely high or
low total costs. Given the small sample size for some of the sub-populations, those outliers may
have a significant effect on the average costs per group.
The cost estimates are based on capitation and fee-for-service (FFS) payment information.
Capitation summaries included regular monthly payments made by DHMH to a member’s
managed care organization (MCO) and ―kick‖ payments made to a MCO for births and hepatitis
C treatments. FFS payments were stratified into the following six reporting categories: inpatient,
outpatient, professional fees, and all other costs (e.g., pharmacy, dental, long-term care, and
home health).
Table 16 presents the average health care costs for hospital inpatient admissions by treatment
group for CYs 2013 through 2015. Average hospital inpatient health costs for participants in the
Health Homes study group decreased from CY 2013 to CY 2015, while costs for participants in
the comparison group increased during this study period. Total inpatient costs for the study
group decreased slightly from $10,884 in CY 2013 to $10,125 in CY 2015, while total inpatient
costs for the comparison group increased from $10,741 in CY 2013 to $11,821 in CY 2015. This
trend was not observed among dually-eligible Health Home participants, whose average hospital
inpatient health care costs increased between CY 2013 and CY 2015.
8 While the amounts charged for health care services are available on FFS claims, managed care encounters do not
list payment amounts reliably. Costs for health care services received by participants covered by a Medicaid
managed care organization are estimated through an imputation mythology.
31
Table 16. Average Hospital Inpatient Health Care Costs by Treatment Group and Provider Type, CY 2013‐ CY 2015
Average Hospital Inpatient Health Costs
Health Home Study Group Comparison Group
Provider Type
CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP $18,049 $32,974 $15,038 $13,570 $17,940 $16,365
MTS $26,138 $19,223 $23,463 $23,715 $27,201 $26,174
PRP $17,082 $17,540 $19,411 $20,254 $19,969 $21,226
Total $19,228 $19,473 $17,565 $19,925 $20,274 $21,189
Duals OTP $1,404 $1,479 $1,820 $5,854 $2,621 $5,022
MTS $1,885 $3,438 $1,564 $3,259 $5,849 $1,936
PRP $1,870 $2,114 $2,363 $2,044 $2,618 $2,028
Total $1,908 $2,231 $2,262 $2,285 $2,785 $2,198
All Participants
OTP $12,501 $23,975 $12,394 $10,426 $13,051 $12,584
MTS $17,605 $14,762 $15,642 $14,510 $18,660 $16,231
PRP $8,470 $8,190 $8,691 $10,578 $11,052 $11,411
Grand Total $10,884 $10,882 $10,125 $10,741 $11,670 $11,822
Table 17 presents the average health care costs for outpatient health care by treatment group for
CY 2013 through CY 2015. Average outpatient health costs increased across the study period for
both the Health Home study group and the comparison group. Outpatient health costs were
slightly higher for participants in the study group for both duals and non-duals. Among non-
duals, MTS participants had higher average outpatient costs than OTP and PRP program
participants did.
32
Table 17. Average Outpatient Health Care Costs by Treatment Group and Provider Type, CY 2013‐ CY 2015
Average Outpatient Health Costs
Health Home Study Group Comparison Group
Provider Type
CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP $2,591 $1,855 $4,424 $1,497 $3,266 $3,743
MTS $7,113 $9,166 $8,938 $4,133 $5,770 $6,905
PRP $3,818 $4,458 $5,167 $3,839 $3,907 $4,401
Total $4,226 $4,724 $5,860 $3,595 $3,936 $4,465
Duals OTP $529 $941 $634 $1,993 $1,502 $1,122
MTS $652 $723 $807 $1,143 $932 $1,245
PRP $1,312 $1,100 $1,110 $1,064 $1,147 $973
Total $1,347 $1,199 $1,196 $1,118 $1,162 $990
All Participants
OTP $2,137 $1,634 $3,571 $1,652 $2,706 $2,937
MTS $3,787 $4,301 $4,701 $2,445 $2,979 $4,027
PRP $2,183 $2,172 $2,381 $2,186 $2,270 $2,373
Grand Total $2,641 $2,777 $3,262 $2,175 $2,352 $2,495
Table 18 presents the average health care costs for professional services (including costs for
physicians and specialists) by treatment group for CY 2013 through CY 2015. Average health
care costs for professional services increased among the participants in the Health Home study
group between CY 2013 and CY 2015, while costs decreased among the comparison group
during the same time period. Study group participants had higher average professional health
care costs than participants in the comparison group did across the study period, regardless of
dual-eligibility status or provider type.
33
Table 18. Average Professional Services Health Care Costs by Treatment Group and Provider Type, CY 2013‐ CY 2015
Average Professional Health Costs
Health Home Study Group Comparison Group
Provider Type
CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP $4,545 $4,129 $7,623 $2,047 $2,290 $5,024
MTS $15,781 $17,576 $18,247 $9,298 $9,427 $9,209
PRP $19,932 $19,892 $19,886 $8,926 $8,081 $8,291
Total $13,478 $13,815 $15,202 $8,183 $7,399 $7,698
Duals OTP $6,014 $7,135 $8,207 $4,355 $3,952 $5,080
MTS $15,657 $16,362 $16,894 $10,255 $10,091 $9,337
PRP $23,144 $23,201 $22,579 $11,111 $10,052 $9,304
Total $20,461 $20,942 $20,795 $10,883 $9,845 $9,179
All Participants
OTP $4,715 $4,435 $7,677 $2,354 $2,536 $5,030
MTS $15,743 $17,169 $17,791 $9,617 $9,680 $9,257
PRP $21,459 $21,516 $21,232 $9,767 $8,882 $8,710
Grand Total $16,161 $16,539 $17,353 $9,165 $8,333 $8,252
34
Table 19 presents the average health care costs for all other services by treatment group for CYs
2013 through 2015. Other services include prescriptions, long-term care, home health, dental,
capitation payments, kick payments, and any other service incurred. Among all participants,
other services health costs increased from CY 2013 to CY 2015. As a whole, participants in the
Health Home study group had lower other services health costs than the comparison group did.
Other services health costs for dually-eligible Health Home participants were less than half of
costs for dually-eligible participants in the comparison group in CY 2014 and CY 2015.
Table 19. Average Other Services Health Care Costs by Treatment Group and Provider Type, CY 2013‐ CY 2015
Average Other Services Health Costs
Health Home Study Group Comparison Group
Provider Type
CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP $13,876 $18,421 $19,585 $12,855 $16,770 $18,710
MTS $19,206 $23,840 $24,120 $16,654 $18,639 $18,619
PRP $16,522 $19,123 $19,966 $15,661 $18,945 $19,400
Total $15,528 $18,840 $19,760 $15,124 $18,417 $19,115
Duals OTP $4,319 $3,631 $5,407 $9,857 $9,547 $10,115
MTS $6,583 $10,852 $8,164 $20,763 $16,446 $19,510
PRP $8,193 $8,274 $9,296 $15,625 $18,726 $22,624
Total $8,380 $8,494 $9,063 $15,705 $18,255 $22,132
All Participants
OTP $13,626 $17,994 $19,248 $12,758 $16,426 $18,405
MTS $17,662 $22,266 $22,760 $17,176 $18,222 $18,780
PRP $15,077 $16,978 $17,812 $15,655 $18,906 $19,965
Grand Total $14,669 $17,429 $18,231 $15,205 $18,392 $19,580
35
Table 20 presents the total average health care costs by treatment group for CYs 2013 through
2015. Total average health care costs increased across the study period for participants in the
study and comparison groups. Participants in the Health Home study group had higher total
average health care costs than participants in the comparison group, with the exception of dually-
eligible participants receiving services through an OTP provider. These participants had lower
average total health costs than their non-Health Homes counterparts.
Table 20. Average Total Health Care Costs per Person by Treatment Group and Provider Type, CY 2013‐ CY 2015
Average Total Health Care Costs
Health Home Study Group Comparison Group
Provider Type
CY 2013 CY 2014 CY 2015 CY 2013 CY 2014 CY 2015
Non-Duals OTP $18,796 $24,758 $29,924 $14,861 $20,322 $25,471
MTS $44,021 $48,709 $49,693 $31,703 $33,684 $34,715
PRP $39,845 $42,284 $43,444 $27,246 $29,696 $30,750
Total $31,425 $36,032 $39,007 $25,189 $28,097 $29,804
Duals OTP $ 7,966 $ 9,243 $10,417 $11,565 $10,237 $11,247
MTS $18,875 $20,551 $19,361 $18,296 $18,272 $17,873
PRP $26,125 $26,355 $26,030 $17,092 $17,426 $17,647
Total $23,647 $24,344 $24,395 $17,029 $17,266 $17,479
All Participants
OTP $17,842 $23,317 $28,108 $14,590 $19,302 $24,086
MTS $36,406 $39,323 $39,697 $27,292 $28,017 $28,630
PRP $33,337 $34,488 $34,762 $23,445 $24,963 $25,614
Grand Total $28,625 $31,675 $33,416 $22,439 $24,306 $25,430
36
Section 5. Health Home Regression Analysis Results
This section of the report presents the results of regression analyses that tested the effects of
Health Home participation on health care utilization, care quality, and cost outcomes.
Multivariate regression analysis is a method for understanding the relationship between an
outcome of interest (e.g., inpatient admission or emergency department visit) and independent
variables expected to affect that dependent variable (e.g. participation in a program expected to
improve health outcomes). Lacking a randomized experimental design for this study, it is
essential to control for differences in characteristics of the patients in the study group and control
group by adding independent variables to measure those characteristics in the regression model.
Explicitly including data on measures (e.g. age, health status, and program) that may influence
the dependent variable allows the model to rigorously assess the effect of the primary
independent variable(s) of interest (e.g. emergency department use).9
The regression method used to analyze the effects of participation in the Health Home program
was difference-in-differences. Difference-in-differences is a quasi-experimental method used to
examine the effects of program interventions in which the outcomes hypothesized to be impacted
by program participation are obtained for both the treatment and comparison group before and
after program implementation. The tables and figures below present the unadjusted average CY
2013 and CY 2015 results as well as the difference-in-differences regression results for the study
and comparison groups. A detailed description of the methodology and regression models used
for the various outcomes is included in Appendix B.
Emergency Department Utilization
Figure 9 presents the unadjusted average number of ED visits per year. The average number of
ED visits did not change significantly between CY 2013 and CY 2015 for either group and rose
slightly for the study group. In CY 2013, the average number of ED visits for the study group
was 2.20 and the average number of ED visits per year was 2.30 for the comparison group. In
CY 2015, the average number of ED visits per year was 2.32 for the study group and 2.30 for the
comparison group.
9 Rubinfeld, D. L. (2000). Reference guide on multiple regression. Reference manual on scientific
evidence, 179, 425-469.
37
Figure 9. Average Number of ED Visits, by Treatment Group, CY 2013 and CY 2015
Table 21 presents the negative binomial regression results modeling the effect of participation in
the Health Home program on counts of ED visits. Having high or very-high co-morbidity, being
a younger adult, being Black, living in the Baltimore Metropolitan Region, and visiting a MTS
provider were each associated with higher counts of ED visits. The estimated Incidence Rate
Ratio (IRR) for the HH variable is 0.89. This estimate suggests that, holding all other variables
constant, if a person in the study group is expected to go the ED 2 times a year, a Health Home
participant is expected to go to the ED 2*(0.89) = 1.78 times a year at baseline. The IRR estimate
of 1.03 for the POST variable indicates that both the study and comparison group had 3 percent
more visits to the ED in CY 2015 than they did in CY 2013. The IRR estimate of 1.07 for the
HH_POST variable suggests that participating in the HH program is related to having 7 percent
more ED visits between CY 2013 and CY 2015, holding all other variables constant.
Table 21. Results of Regression on Counts of ED Visits, CY 2013 and CY 2015 OUTCOME: Counts of ED Visits
Negative Binomial Regression Model
Independent Variable Name
Independent Variable Description Incidence Rate Ratio Estimates
HH Health Home Program Indicator 0.89 **
POST POST Time Period Indicator (CY2015) 1.03 **
HH_POST HH*POST Interaction Term 1.07 **
RACE/ETHNICITY Black 1.24 **
SEX Female 1.09 **
AGE GROUP 40 to 64 0.65 **
38
ACG COMORBIDITY
High Comorbidity 1.99 **
Very-High Comorbidity 3.89 **
REGION Baltimore Metropolitan Region 1.16 **
Montgomery and Prince George's Counties 0.65 **
Western Maryland 0.97
TYPE OF PROVIDER
Visited an OTP Provider During CY2012 0.84 **
Visited an MTS Provider During CY2012 1.45 **
Visited a PRP Provider During CY2012 0.73 **
The asterisks are * for 95% statistical significance and ** for 99%.
Inpatient Hospital Admissions
Figure 10 presents the unadjusted average number of inpatient hospital admissions per year. The
average number of inpatient hospital admissions per year went down for the study group between
CY 2013 and CY 2015, while remaining the same for the comparison group for the same period.
In CY 2013, the average number of inpatient hospital admissions was 0.47 for the study group
and 0.42 for the comparison group. In CY 2015, the average number of inpatient hospital
admissions per year for the study group dropped slightly to 0.44 and the average number of
inpatient hospital admissions per year for the comparison group remained at 0.42.
Figure 10. Average Number of Inpatient Hospital Admissions, by Treatment Group, CY 2013 and CY 2015
Table 22 presents the regression results modeling the effect of participation in the Health Home
program on counts of inpatient hospital admissions. Having high or very-high co-morbidity,
being a younger adult, living in Baltimore, Montgomery, or Prince George’s County, and visiting
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
CY2013 CY2015
Study Group
Comparison Group
39
an MTS provider were each associated with higher inpatient hospital admissions counts. The
IRR estimate of 1.07 for the HH variable indicates that those in the Health Home program have 7
percent more inpatient hospital admissions than those in the comparison group. The IRR estimate
of 1.07 for the POST variable indicates that the study and comparison group have 7 percent more
inpatient hospital admissions in CY 2015 than they did in CY 2013. The IRR estimate of 0.93 for
the HH_POST variable suggests that participating in the HH program is related to having 7
percent fewer inpatient hospital admissions between CY 2013 and CY 2015, holding all other
variables constant.
Table 22. Results of Regression on Counts of Inpatient Hospital Admissions, CY 2013 and CY 2015
OUTCOME: Counts of Inpatient Hospital Admissions
Negative Binomial Regression Model
Independent Variable Name
Independent Variable Description Incident Rate
Ratio Estimates
HH Health Home Program Indicator 1.07 **
POST (CY15) POST Time Period Indicator (CY2015) 1.07 **
HH_POST HH*POST Interaction Term 0.93 **
RACE/ETHNICITY Black 1.04 **
SEX Female 1.00 **
AGE GROUP 40 to 64 0.74 **
ACG COMORBIDITY
High Comorbidity 1.77 **
Very-High Comorbidity 3.52 **
REGION Baltimore Metropolitan Region 1.41 **
Montgomery and Prince George's Counties 1.30 **
Western Maryland 1.20 **
TYPE OF PROVIDER
Visited an OTP Provider During CY2012 0.80 **
Visited an MTS Provider During CY2012 1.72 **
Visited a PRP Provider During CY2012 0.76 **
The asterisks are * for 95% statistical significance and ** for 99%.
Non-Emergent Emergency Department Visits
Figure 11 presents the unadjusted average number of non-emergent ED visits per year. The
number of ED visits classified as non-emergent dropped for both the comparison and study
groups between CY 2013 and CY 2015. In CY 2013, the average number of non-emergent ED
visits was 1.53 for the study group and 1.59 for the comparison group. In CY 2015, the average
number of non-emergent ED visits per year was 0.84 for the study group and 0.86 for the
comparison group.
40
Figure 11. Average Number of Non-Emergent ED Visits, by Treatment Group, CY 2013 and CY 2015
Table 23 presents the regression results modeling the effect of participating in the Health Home
program on counts of non-emergent ED visits. Having high or very-high co-morbidity, being
Black, being a younger adult, and visiting an MTS provider were each associated with higher
non-emergent ED visit counts. People living in Montgomery and Prince George’s Counties and
those that visited a PRP provider were less likely to have a non-emergent ED visit. The IRR
estimate for the HH variable indicates that those in the Health Home program have 12 percent
fewer non-emergent ED visits than those in the comparison group. The IRR estimate of 0.58 for
the POST variable indicates that the study and comparison group had 42 percent fewer non-
emergent ED visits in CY 2015 than they did in CY 2013. The IRR estimate of 1.02 for the
HH_POST variable suggests that participating in the HH program is related to having 2 percent
more non-emergent ED visits between CY 2013 and CY 2015, holding all other variables
constant.
Table 23. Results of Regression on Counts of Non-Emergent ED Visits, CY 2013 and CY 2015
OUTCOME: Counts of Non-Emergent ED Visits
Negative Binomial Regression Model
Independent Variable Name
Independent Variable Description Incident Rate
Ratio Estimates
HH Health Home Program Indicator 0.88 **
POST (CY15) POST Time Period Indicator (CY2015) 0.58 **
HH_POST HH*POST Interaction Term 1.02 **
RACE/ETHNICITY Black 1.37 **
SEX Female 1.24 **
41
AGE GROUP 40 to 64 0.73 **
ACG COMORBIDITY
High Comorbidity 1.79 **
Very-High Comorbidity 2.82 **
REGION Baltimore Metropolitan Region 1.08 **
Montgomery and Prince George's Counties 0.53 **
Western Maryland 0.99 **
TYPE OF PROVIDER
Visited an OTP Provider During CY2012 0.91 **
Visited an MTS Provider During CY2012 1.18 **
Visited a PRP Provider During CY2012 0.78 **
The asterisks are * for 95% statistical significance and ** for 99%.
Figure 12 presents the unadjusted percentages of participants with at least one avoidable hospital
admission (PQI) per year. The percent of the population with an avoidable hospital admission
increased slightly for both the study and comparison groups. In CY 2013, 1.91 percent of the
study group had a PQI and 1.79 percent of the comparison group had a PQI. In CY 2015, 2.31
percent of the study group had a PQI and 1.98 percent of the comparison group had a PQI.
Avoidable Hospital Admissions
Figure 12. Percentage of Participants with at Least One Avoidable Hospital Admission, by Treatment Group, CY 2013 and CY 2015
Table 24 presents the regression results modeling the effect of participation in the Health Home
program on the likelihood of having at least one PQI per year. Having a high or very-high co-
morbidity, being female, and/or being an older adult were each associated with increased
likelihoods of having a PQI. People who visited a PRP were about half as likely to have a PQI as
those that did not. The IRR estimates for the HH, POST, and HH_POST variables indicate that
42
there is no significant relationship between joining the program, time passing from CY 2013 and
CY 2015, or participating in the program over time on the likelihood of an avoidable admission.
Table 24. Results of Regression on the Likelihood of an Avoidable Hospital Admission, CY 2013 and CY 2015
OUTCOME: Likelihood of an Avoidable Hospital Admission
Logistic Regression Model
Independent Variable Name
Independent Variable Description Odds Ratio Estimates
HH Health Home Program Indicator 0.93
POST (CY15) POST Time Period Indicator (CY2015) 1.11
HH_POST HH*POST Interaction Term 1.10
RACE/ETHNICITY Black 1.19
SEX Female 1.47 **
AGE GROUP 40 to 64 1.74 **
ACG COMORBIDITY
High Comorbidity 1.78 **
Very-High Comorbidity 5.76 **
REGION
Baltimore Metropolitan Region 1.43
Montgomery and Prince George's Counties 1.36
Western Maryland 1.59
TYPE OF PROVIDER
Visited an OTP Provider During CY2012 1.24
Visited an MTS Provider During CY2012 1.12
Visited a PRP Provider During CY2012 0.49 **
The asterisks are * for 95% statistical significance and ** for 99%.
30-Day Hospital Readmissions
Figure 13 presents the unadjusted percentages of participants with at least one 30-day hospital
readmission per year. The percent of the population with a 30-day hospital readmission increased
slightly for the study group while dropping slightly for the comparison group. In CY 2013, 4.13
percent of the study group had a 30-day hospital readmission, and 4.22 percent of the
comparison group had a 30-day hospital readmission. In CY 2015, 4.41 percent of the study
group had a 30-day hospital readmission, and 3.74 percent of the comparison group had a 30-day
hospital readmission.
43
Figure 13. Percentage of Participants with at Least One 30-Day Hospital Readmission, by Treatment Group, CY 2013 and CY 2015
Table 25 presents the regression results modeling the effect of participating in the Health Home
program on the likelihood of having at least one 30-day hospital readmission per year. Having a
high or very-high co-morbidity, being Black and/or a younger adult, and visiting an MTS
provider were each associated with increased likelihoods of having a 30-day hospital
readmission. The odds ratio estimates for the HH, POST, and HH_POST variables indicate that
there is no significant relationship between joining the program, time passing from CY 2013 and
CY 2015, or participating in the Health Home program between CY 2013 and CY 2015 on the
likelihood of an avoidable admission.
Table 25. Results of Regression on the Likelihood of a 30-day Hospital Readmission, CY 2013 and CY 2015
OUTCOME: Likelihood of a 30-Day Hospital Readmission
Logistic Regression Model
Independent Variable Name
Independent Variable Description Odds Ratio Estimates
HH Health Home Program Indicator 0.91
POST (CY15) POST Time Period Indicator (CY2015) 0.88
HH_POST HH*POST Interaction Term 1.22
RACE/ETHNICITY Black 1.30 **
SEX Female 0.96
AGE GROUP 40 to 64 0.66 **
ACG COMORBIDITY
High Comorbidity 1.72 **
Very-High Comorbidity 3.79 **
REGION
Baltimore Metropolitan Region 1.56 **
Montgomery and Prince George's Counties 1.37
Western Maryland 1.31
TYPE OF Visited an OTP Provider During CY2012 0.76
0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
CY2013 CY2015
Study Group
Comparison Group
44
PROVIDER Visited an MTS Provider During CY2012 1.89 **
Visited a PRP Provider During CY2012 0.80
The asterisks are * for 95% statistical significance and ** for 99%.
Cost Outcomes
Figure 14 presents the unadjusted average annual total health care costs per year. The average
total costs increased for both the study and comparison groups, but the total annual costs for the
study group started higher than the comparison group and remained higher. In CY 2013, the
average total health care costs of the study group were $28,148, and the average total health care
costs were $22,109 for the comparison group. In CY 2015, the average total health care costs of
the study group increased to $33,174, and the average total health care costs of the comparison
group increased to $24,911.
Figure 14. Average Total Annual Health Care Costs, by Treatment Group, CY 2013 and CY 2015
Table 26 presents the regression results modeling the effect of participating in the Health Home
program on the log of the total annual health care costs. Higher levels of comorbidities, visiting
an MTS and/or PRP provider, and living in the Baltimore Metropolitan, Montgomery and Prince
George’s County, or Western regions were each associated with increases in total annual average
health care costs. The parameter estimate for the HH variable indicates that those in the Health
Home program have 50 percent higher annual health care costs than those in the comparison
group. The parameter estimate for the HH_POST variable suggests that participating in the
Health Home program is related to having 24 percent higher total annual health care costs
between CY 2013 and CY 2015, holding all other variables constant.
$0
$5,000
$10,000
$15,000
$20,000
$25,000
$30,000
$35,000
CY2013 CY2015
Study Group
Comparison Group
45
Table 26. Results of Regression on Total Annual Health Care Costs, CY 2013 and CY 2015 OUTCOME: Log of the Total Annual Health Care Costs
Linear Regression Model
Independent Variable Name
Independent Variable Description Parameter Estimates
HH Health Home Program Indicator 0.50 **
POST (CY15) POST Time Period Indicator (CY2015) -0.05
HH_POST HH*POST Interaction Term 0.24 **
RACE/ETHNICITY Black -0.02
SEX Female -0.01
AGE GROUP 40 to 64 0.08 **
ACG COMORBIDITY
High Comorbidity 0.24 **
Very-High Comorbidity 0.54 **
REGION Baltimore Metropolitan Region 0.22 **
Montgomery and Prince George's Counties 0.25 **
Western Maryland 0.10 *
TYPE OF PROVIDER
Visited an OTP Provider During CY2012 0.01
Visited an MTS Provider During CY2012 0.56 **
Visited a PRP Provider During CY2012 0.43 **
The asterisks are * for 95% statistical significance and ** for 99%.
Limitations
The data presented in this report were current as of October 31, 2016. Typically, MMIS2 data are
not considered complete until 12 months have passed for adjudication of FFS claims and 6
months for submission of managed care encounters. Therefore, measures based on MMIS2 data,
particularly for recently occurring time periods, should be considered preliminary.
The results presented in this report should be interpreted with caution, as sufficient time has not
passed since the implementation of the program to detect meaningful and sustained differences
in long-term health outcomes. Further, increases and decreases in care utilization observed when
comparing the Health Home study and comparison groups were small. Due to the limited data
available, it is difficult to discern whether fluctuations in utilization can be attributed exclusively
to participation in a Health Home or were driven by other causes. The Health Homes often have
few participants, especially when limiting the analysis to people who were continuously enrolled
for a full calendar year. In addition, the results of this analysis are not generalizable to the
Medicaid population at large.
The cost information provided does not exclude outliers with extremely high or low total costs
and is not adjusted to account for changes in the fees paid by Medicaid for the same service
46
received at different points in time. The final evaluation may include other components such as
analysis within different time intervals, participants’ diagnoses and health services history, and
information about the Health Home program implementation.
47
Conclusion
Health Homes are intended to improve health outcomes for individuals with chronic conditions
by providing patients with an enhanced level of care management and care coordination. The
Maryland Health Home program is aimed at Medicaid participants with either a SPMI and/or
an opioid SUD and risk of additional chronic conditions due to tobacco, alcohol, or other non-
opioid SUD, and children with SED. The information presented in this report provides
evidence that Health Homes successfully tie this extremely vulnerable population to social and
somatic care services and improve their access to preventive care.
The results of this preliminary analysis suggest that Health Home participants had a strong
demand for the Health Home social services, such as care coordination and health promotion.
When comparing the study group to a comparison group of Medicaid participants with similar
characteristics, the results show mixed results in the overall trends for the health care
utilization and outcomes measures for each group. For example, the Health Home study group
had larger increases in rates of ambulatory care between CY 2013 and CY 2015 than the
comparison group did.
Overall, both the descriptive and regression analyses included in this report do not offer
conclusive evidence that Health Home participants experience better health care utilization,
quality, and cost outcomes than their control group counterparts. Within the descriptive results,
although the comparison group’s overall utilization of services was often higher than that of the
study group, the comparison group experienced more decreases in inpatient stays, ED visits,
30-day all-cause hospital readmissions, and avoidable ED visits. However, one outcome,
inpatient admissions, had recurring positive results. Despite a higher overall rate of inpatient
admissions, the descriptive results suggest that the average length of stay for those hospitalized
was lower for certain populations within the study group than the comparison group.
Additionally, the regression analysis indicates that there was a statistically significantly larger
decrease in the frequency of inpatient visits for Health Home participants relative to that of the
comparison group between CY 2013 and CY 2015, holding all other variables constant.
A complete evaluation of this program will be completed once more time has passed for the
anticipated long-term outcomes to present themselves.
48
Appendix A: Evaluation Cohort Selection
To select a group of Medicaid participants to use as an appropriate comparison for the health
outcomes of participants in the Health Home program, the team first identified adults that had
seen a PRP, OTP, or MTS provider and were continuously enrolled in Medicaid CY 2013 to CY
2015. Once the selection of potential matches was finished, the team implemented propensity
score matching—a statistical technique that attempts to select a group of controls with which to
compare the study population and minimize potential bias when comparing program-related
outcomes. Propensity score matching creates a sample of participants and non-participants that
are comparable across a set of independent characteristics theorized to have an effect on the
outcomes of interest. By doing this, the groups are constructed to have a relatively similar
likelihood of joining the study. For this analysis, the team created propensity scores by
estimating a regression of the likelihood of joining the program on the following independent
characteristics: geographic region of residence, age, race/ethnicity, gender, ACG co-morbidity
grouping, dual, eligibility, and type of Health Home provider seen. The result was a one-to-one
match between the study and comparison groups. Each member of comparison group can only be
matched to one participant in the study group.
Table 27. Number of Health Home and Comparison Group Participants
Selection Criteria Health Home Participants
Medicaid Participants
Full Group 7,905 1,226,105
Adults (aged 18‐64) that had seen a PRP, OTP, or MTS provider and were continuously enrolled in Medicaid CY 2013 to CY 2015
5,468
16,209
Found an appropriate match via the propensity score selection process
3,290 3,290
To develop estimates of the outcomes of interest, the team used the generalized linear model
procedure. The procedure takes into account the differences between the two groups, including
their outcome variances, participation in the study versus comparison group, as well as the
individual’s propensity score.
Because of the propensity score method used to select the evaluation cohort, this analysis should
not be considered to be generalizable to the Medicaid population at large or to all participants in
the Health Home program. The people in the comparison group are only those that sought out
care in a PRP, OTP, or MTS facility, as well as meeting the other criteria to match to the study
population of interest. Furthermore, developing the group to be studied required reducing the
sample by removing cases at the high and low ends of the distributions of the estimates
propensity scores.
49
Appendix B: Regression Methods
Model Selection
Multivariate regression analysis is a method for understanding the relationship between a
dependent variable (i.e. a specific measured outcome of interest; e.g. inpatient admission or
emergency department visit) and independent variables expected to affect the dependent
variable. A regression estimates the effects those independent variables have on that outcome by
constructing an equation that minimizes differences between the actual values and the values that
fit on the curve or line described by the equation. Lacking a randomized experimental design for
this study, it is essential to control for systematic differences in characteristics of the patients in
the study group and control group by adding independent variables to measure those
characteristics in the regression model. Explicitly including data on measures (e.g. age, health
status, and program) that may influence the dependent variable allows the model to control for
these factors and rigorously assess the effect of the primary independent variable(s) of interest on
the outcome.10
Difference-in-differences (DD) is a quasi-experimental method used to examine the effects of
program interventions, such as the Health Homes program. To conduct a DD evaluation, the
outcomes hypothesized to be impacted by program participation are obtained for both the
treatment and comparison group, before and after program implementation. For this analysis, the
outcomes of interest are health care utilization, quality, and cost outcomes. The DD method then
estimates whether changes in those dependent variables, between the pre- and post-intervention
periods for those that participated in the program, are statistically different from the changes over
that time for those that did not participate in the Health Homes program. The benefit of using DD
is that each group’s baseline outcome serves as the group’s own control to account for
unobservable but fixed characteristics. A key assumption for DD is that the outcome in the study
and comparison groups would change over time in the absence of the treatment such that the
baseline difference between groups is a good estimate of what the post-baseline difference would
have been. Using propensity score matching to select the comparison group helps to strengthen
this assumption (see Appendix A).
In the DD analysis developed for this evaluation, the regression equation has the following
structure:
Yi = β0 + β1(HH) + β2(POST) + β3 (HH *POST) + βk (Other Controls) + εi (Error)
The values of β are the coefficients of each variable and have the following interpretations:
10 Rubinfeld, D. L. (2000). Reference guide on multiple regression. Reference manual on scientific
evidence, 179, 425-469.
50
β1 measures the treatment group effect (i.e. average differences between treatment and control).
HH = 1 if the observation is a person participating in the Health Home program. HH = 0 if the
observation is a person not participating in the Health Home program.
β2 accounts for the time periods. POST = 1 if the observation occurred during CY 2015. POST =
0 if the observation occurred during CY 2013.
β3 is the coefficient on HH_POST, an interaction term, i.e., the product of the HH and POST
categorical variables. HH_POST = 1 if the observation is a person participating in the Health
Home program and occurred during CY 2015. Otherwise, HH_POST = 0.
HH_POST is the most important term in the DD regression. It represents the true effect of
participating in the intervention, because it controls for changes over time that are seen in both
the study and comparison group. This removes biases in second period comparisons between the
study and comparison group that could be the result of permanent differences between those
groups, as well as biases from comparisons over time that could be the result of trends.
Βk is a set of coefficients for the effects of other control variables.
β0 and εi are, respectively, the estimates of the equation’s intercept (when all the variables are
zero) and the equation’s errors (differences between the actual values and the values predicted by
the equation). These coefficients are not included in the tables below, as they do not assist with
interpretation of the outcomes.
Interpretation of Tables
Different regression models were selected according to the types of statistical distributions of the
outcome variables.
Logistic regression models were used for the likelihood of any:
Avoidable hospital admissions, and
30-day hospital readmissions.
Negative binomial regression models were used for the counts of:
ED visits,
Inpatient hospital admissions, and
Non-emergent ED visits.
A linear regression model was used for:
51
Total health care costs, after the costs were converted to logarithms.
In addition, analysts tested alternatives for the selection of the explanatory variables used and
years of pre- and post- intervention to include in the models.
The numbers in the tables describing logistic regression results are odds ratio estimates. Odds
ratios represent the change in the probability of the outcome of interest resulting from a change
in the independent variable of one unit. For example, the odds ratio estimate for a value of ―1‖ on
the HH variable for having a non-avoidable hospital admission regression is 0.93 or about 7
percent less likely. An odds ratio less than one means the outcome is less likely; an odds ratio
greater than one means an outcome is more likely.
The coefficients in the tables for a negative binomial regression are interpreted differently than
the logistic regression results. Because the dependent variable is a count of how many times an
outcome occurs over the course of the measurement period, the coefficient is called an Incidence
Rate Ratio (IRR). For example, the model of the frequency of ED visits, the estimated IRR for
the HH variable is 0.89. This regression estimates that, holding all other variables constant, if a
person in the study group is expected to go the ED 2 times a year, a Health Home participant is
expected to go to the ED 2*(0.89) = 1.78 times a year, at baseline. An IRR greater than one
means the outcomes are more frequent; an IRR less than one means the outcomes are less
frequent.
For the total annual health care costs regression, the cost outcome variable is a continuous
measure rather than discrete counts. After converting the costs into logarithms, the parameter
estimates for that model represent the percent change in the value of the outcome expected for
every one-unit change in the explanatory variable. For example, the parameter estimate for the
HH variable is 0.50 in the total annual health care costs regression. This regression estimates
that, holding all other variables constant, those in the Health Home group are expected to have
50% higher total costs than those in the comparison group at baseline. Negative regression
coefficients predict lower costs, while positive coefficients predict higher costs.
52
References
American Academy of Family Physicians (AAFP), American Academy of Pediatrics (AAP),
American College of Physicians (ACP), American Osteopathic Association (AOA). (2007,
March). Joint Principles of the Patient-Centered Medical Home. Retrieved from
http://www.aafp.org/dam/AAFP/documents/practice_management/pcmh/initiatives/PCMHJo
int.pdf
Bentler, S., Damiano, P., Momany, E., McInroy, B., Robinson, E., & Pooley, M. (2015).
Evaluation of Iowa Health and Wellness Plan member experiences in the first year. Iowa
City, IA: University of Iowa Public Policy Center. Retrieved from
http://ppc.uiowa.edu/sites/default/files/ihh_report_july_2015.pdf
Billings, J., Parikh, N., & Mijanovich, T. (2000, November). Emergency Room Use: The New
York Story, New York, The Commonwealth Fund. Retrieved from
http://www.commonwealthfund.org/usr_doc/billings_nystory.pdf
The Centers for Medicare & Medicaid Services. (2010, November). Letter to State Medicaid
Director and State Health Official, SMDL#10-024, ACA#12. Retrieved from
http://downloads.cms.gov/cmsgov/archived-downloads/SMDL/downloads/SMD10024.pdf
The Centers for Medicare & Medicaid Services. (2013, September). Maryland Health Home
State Plan Amendment. Retrieved from https://www.medicaid.gov/state-resource-
center/medicaid-state-technical-assistance/health-homes-technical-
assistance/downloads/maryland-spa-13-15.pdf
The Centers for Medicare & Medicaid Services. (2014, March). Core set of health care quality
measures for Medicaid Health Home programs. Retrieved from
https://www.medicaid.gov/state-resource-center/medicaid-state-technical-assistance/health-
homes-technical-assistance/downloads/health-home-core-set-manual.pdf
The Centers for Medicare & Medicaid Services. (2014, October). Behavioral Health Services.
Retrieved from https://www.medicaid.gov/medicaid/benefits/bhs/index.html
The Centers for Medicare & Medicaid Services. (2015, May). Medicaid Health Homes: An
Overview. Retrieved from https://www.medicaid.gov/state-resource-center/medicaid-state-
technical-assistance/health-homes-technical-assistance/downloads/medicaid-health-homes-
overview.pdf
The Centers for Medicare & Medicaid Services. (2016, July). State-by-State Health Home State
Plan Amendment Matrix. Retrieved from https://www.medicaid.gov/state-resource-
center/medicaid-state-technical-assistance/health-homes-technical-
assistance/downloads/state-hh-spa-at-a-glance-matrix-jul-2016.pdf
53
The Centers for Medicare & Medicaid Services. (2016, September). Approved Medicaid Health
Home State Plan Amendments. Retrieved from https://www.medicaid.gov/state-resource-
center/medicaid-state-technical-assistance/health-homes-technical-assistance/downloads/hh-
map_v55.pdf
Department of Health and Human Services (2014). Interim Report to Congress on the Medicaid
Health Home State Plan Option. Retrieved from
https://www.medicaid.gov/medicaid/ltss/downloads/health-homes/medicaid-health-home-
state-plan-option.pdf
Department of Mental Health and MO HealthNet (2013, June). Progress Report: Missouri
CMHC Healthcare homes. Retrieved from
http://dmh.mo.gov/docs/mentalillness/prnov13.pdf
Department of Health and Mental Hygiene (2016, October). Maryland Medicaid Chronic Health
Homes Quarterly Report: July to September 2016. Retrieved from
https://mmcp.dhmh.maryland.gov/SiteAssets/SitePages/Health%20Home%20Program%20U
pdates/Quarterly%20Health%20Home%20Report-%20July%20-
%20September%202016.pdf
Dickey, B., Normand, S. L. T., Weiss, R. D., Drake, R. E., & Azeni, H. (2002, July). Medical
morbidity, mental illness, and substance use disorders. Psychiatric Services, 53(7), 861-867.
Retrieved from http://ps.psychiatryonline.org/doi/pdf/10.1176/appi.ps.53.7.861
General Assembly of Maryland (2015, April). Joint Chairmen’s Report. Report on the Fiscal
2016 State Operating Budget (HB 70) and the State Capital Budget (HB 71) and Related
Recommendations by the Chairmen of the Senate Budget and Taxation Committee and House
Appropriations Committee. Annapolis, MD. Retrieved from
http://mgaleg.maryland.gov/Pubs/BudgetFiscal/2015rs-budget-docs-jcr.pdf
Higgins, S., Chawla, R., Colombo, C., Snyder, R., & Nigam, S. (2014, March). Medical homes
and cost and utilization among high-risk patients. The American Journal of Managed Care,
20(3), e61-71. Retrieved from http://www.ajmc.com/journals/issue/2014/2014-vol20-
n3/medical-homes-and-cost-and-utilization-among-high-risk-patients
Momany, E.T., Damiano, P.C. & Bentler, S.E. (2014, October). Outcomes for Iowa Medicaid
health home program enrollees: 2011-2013. Retrieved from
http://ppc.uiowa.edu/sites/default/files/healthhomeoutcomesreport.pdf
Rubinfeld, D. L. (2000). Reference guide on multiple regression. Reference manual on scientific
evidence, 179, 425-469Scott, D., & Happell, B. (2011). The high prevalence of poor physical
health and unhealthy lifestyle behaviours in individuals with severe mental illness. Issues in
Mental Health Nursing, 32(9), 589-597.
54
Sia, C., Tonniges, T. F., Osterhus, E., & Taba, S. (2004, May). History of the medical home
concept. Pediatrics, 113(Supplement 4), 1473-1478. Retrieved from
http://pediatrics.aappublications.org/content/pediatrics/113/Supplement_4/1473.full.pdf
Spillman, B. C., Ormond, B. A., and Richardson, E. (2012, December). Evaluation of the
Medicaid Health Home Option for Beneficiaries with Chronic Conditions: Final Annual
Report – Base Year. Washington, DC: The Urban Institute. Retrieved from
https://aspe.hhs.gov/sites/default/files/pdf/76656/HHOption.pdf
West Virginia Department of Health and Human Resources Bureau for Medical Services (2016,
April). Inaugural Year 2014 – 2015 Annual Report: West Virginia Health homes. Retrieved
from
http://www.dhhr.wv.gov/bms/BMSPUB/Documents/HealthHomesAnnualReport%20Final%
20Version.pdf
Wholey, D. R., Finch, M., Shippee, N. D., White, K. M., Christianson, J., Kreiger, R., et al.
(2015, December). Evaluation of the State of Minnesota's Health Care Homes Initiative:
Evaluation Report for Years 2010-2014. Retrieved from
https://www.leg.state.mn.us/docs/2016/mandated/160245.pdf
University of Maryland, Baltimore County Sondheim Hall, 3rd Floor
1000 Hilltop Circle Baltimore, MD 21250
410-455-6854 www.hilltopinstitute.org